Scanning system of the optical readably label for information decoding

BACKGROUND OF THE INVENTION The present invention relates to an improved optically readable label and its reading system, namely, an improved optically readable label, affixed to or printed on a substrate, for writing information within two-dimensional grid data consisting of a set of polygons arranged adjacent to a lattice structure , two different optical characteristics.

Accessory in microfiche

The application is accompanied by a microfiche appendix containing one microfiche and a target test shot and 78 frames from the list of the computer program.

Goods, miscellaneous components, letters, packages, containers, and many other items that are transported or transported, often require identifying information as to their origin, airplane flight number, destination, name, price, product quantity, and so on. Occasionally, reading coded information printed on labels affixed to those products enables you to automate the reading of the digits characterizing the products and their inventory or cash register operation. In other cases, using such encoded labels, automated mail, baggage, etc. may be possible. tracking, sorting, as well as providing necessary information on raw materials and consumables in production. Typically, this type of product is labeled with bar codes, one of which is the Universal Product Code. There are, of course, many other string encoding systems.

Typically, commercial string encoding systems have the disadvantage associated with the constant requirement to increase data density by encoding an increasing amount of information on a potentially smaller label. BanLT 3516 B Dimensions. reducing the label itself, as well as the distances between individual strings in the various strings encoding systems to increase density, did not solve the problem: optical readers with relatively high bar code reading resolution equal to five and less than five miles between adjacent strings become economically unviable due to low label element tolerances, as well as relatively sophisticated optical equipment capable of reading binary bar codes in such dimensions , greatly complicates the production process. In order to adapt to the increasing amount of data, labels with barcodes should be made large in size and not suitable for attachment to small products. Another important factor is that paper labels become very expensive .. A small label is cheaper than a large one; this price becomes an important factor in the case of large volumes of labels.

Alternatives to barcodes are:

circular shapes using radially spaced wedge coded elements, such as U.S. Pat. 3553438, or concentric black and white binary coded rings as in U.S. Pat. 3971917, no. 3916150; a grid of rows and columns, of data squares or rectangles, as disclosed in U.S. Pat. No. 4,286,146, microscopic spots disposed in lattices forming a regularly spaced grid, as in U.S. Pat. 4634850, and densely spaced multicolored data fields from pixels or elements as described in U.S. Pat. 4488679. Some of the encoding systems described and other encoding systems known in the art suffer primarily from insufficient data density, as in the case of encoded round drawings and grids on rectangles and square boxes. Alternatively, in the case of grids consisting of microscopic dots or multicolored elements as discussed above, such systems require special guidance and transport means, limiting their use only under strongly controlled reading conditions.

Due to the size and speed of modern transportation systems using conveyor belts, such as 0.9-1.0 m wide and belt speeds,

2.5 m / sec or even packing at different heights with affixed coded information labels and the need for small, compact and inexpensive labels close to one square inch, create enormous strain on optical and decoding systems that searches for and reads coded information on labels affixed to fast moving packages and similar articles. There is a problem with the optical decompression of the detected label image. In addition, upon detection or identification of a label in a package that is contained in a conveyor system, the label image must be accurately decoded before the next operation begins, and often within a fraction of a second. These problems have made it possible to create a simple, fast, and inexpensive means of signaling the presence of a label with encoded data in the optical field of view of an optical reader comprising the entire length of the conveyor belt. This feature, combined with a high-density data cell, is described in more detail below:

Prefabricated data cells are known in the art: for example, concentric geometric shapes that engage rings, squares, triangles, hexagons, and many variants thereof, as described in U.S. Pat. U.S. Patent No. 3,513,320, 3603720. U.S. Pat. 3693154 and 3801775 also describe the use of symbols consisting of concentric circles as recognition and position indicators affixed to articles themselves where they are optically read. Because the use of two different symbols in such systems is necessary to identify the field data and their position, the complexity of the logical scheme that captures said symbols is increased, as well as the sensitivity of the field data to the data. In this way, using two symbols and failing one of them causes problems in determining the position of the data field and the ability of the operator to reset the information to the field data. The last system uses separate position and orientation tags at opposite ends of the data paths, which contain linear tags of encoded data with limited data sensitivity.

The above systems mainly utilize optical propagation with a sensor capable of generating an output video signal corresponding to a change in the brightness of the light reflected from the data cell and the position and orientation symbols. In such systems, the output video signal, after being quantized, acquires a specific bitmap which can correspond to a given bit sequence. But such systems have the disadvantage of requiring two separate symbols first to familiarize themselves with the image and then to determine its orientation. Also, the necessity of matching the optical sensor output digital signal with a given bit sequence, expressed in symbols and position and orientation, is more likely to cause erroneous readings than in the present invention and the inventive system, since known label detection systems provide inflexible application level recognition.

U.S. Pat. 3553438 shows a circular data cell having a centered recognition application consisting of a series of concentric circles. The AtpaLT 3516 B Knowledge Application validates the tool by an optical sensor that detects a circular label and locates its geometric center, and thus the geometric center of a circular data cell. This is done by a logic scheme that executes the pulse chart recognition, which has a bull eye target configuration configuration. However, in the case of barcodes, the data cell has only limited data capacity and the system requires a second circular spreading process. Using both linear and circular spreading, a system with such a limited data rate would create undesirable system complexity due to a slight increase in data rate with conventional barcodes.

Codes for multiplexing a data-intensive data grid using a plurality of colored high-density dots are described in U.S. Pat. However, the systems described in U.S. Pat. 4488679 requires the use of handheld optical image analyzers that are unable to fully record and decode when the data cells move rapidly on the pack due to high speed conveyor belt transport. Similarly, high-density coding systems using data encoding microscopic pixels (described in U.S. Patent No. 6,348,50) require a special transport device that ensures that the data cells move in the right direction, rather than incidentally as luggage, conveyor belt or in an analogous way. In this way, the coded label must be read tape after tape using a linear reading device coupled to the label transport means for the purpose of decoding the exact information on the label. This patent also states that the position of the card relative to the transducer must be carefully controlled so that it can be read.

Techniques for producing strike coding systems also use multiple colors to avoid optical problems by spreading very fine strings, a stroke code that uses more of two optical characteristics to encode data in a data cell, such as black, gray, and white strokes described in the US. U.S. Pat. 4443694. However, the systems described, although more advanced than the prior art bar code systems, cannot compare to the compactness of the data density described in the present invention.

In view of the shortcomings of the known coding systems, the main object of the invention is to provide a new and improved compact optically readable label with a high density of information.

Another object of the invention is to provide new improved optically readable labels that can encode about 100, well-tolerated, alphanumeric characters per square inch of label.

It is a further object of the invention to provide new improved, high density, optical information readable labels that can be read by an optical sensor when the label is affixed to a luggage or other article transported on a high-speed conveyor system, regardless of its orientation or luggage. optically readable label, height change.

It is also an object of the invention to provide optically readable labels in combination with a decoding system so that the label can be reliably decoded when it is tilted, twisted, wrinkled, partially rubbed, or torn.

It is a further object of the invention to provide methods for locating a label moving at high speed under an optical sensor which will allow decoding of this label with a high degree of data retention.

It is another object of the present invention to provide an improved encoding method for compact, enhanced, optically readable labels with high density of information by dividing the encoded information into high and low priority messages, maintaining a message hierarchy protected from error, and encrypted information integrity.

It is another object of the present invention to provide improved methods and encoding and decoding systems for compact, advanced, high density and correlation capabilities that allow for the reproduction of misspelled or missed high-priority encoded information in optically readable labels.

It is also an object of the present invention to produce inexpensive, optically readable labels by simple print media and to decode them using inexpensive logic circuits.

Other objects and advantages of the invention will become apparent from the following description of the invention.

The present invention consists of an optically readable label for memorizing data encoded in a binary form having a plurality of hexagonal encoded information hexagons arranged adjacent to a two-dimensional military matrix and having at least two different optical characteristics, and methods and apparatus for encoding and decoding such optically readable labels.

In a preferred embodiment of the present invention, the data grid has a substantially square one-square-inch grid with adjacent hexagons forming rows and columns, and a centered recognition application with a geometric center defining a geometric center of the data grid.

The recognition application may be of any geometric shape having optical characteristics that allow the generation of a readily recognizable video signal transmitted by an optical sensor along a linear propagation trajectory passing through the geometric center of the recognition application. In the preferred embodiment, the recognition target is comprised of a plurality of concentric rings with contrasting reflection properties that allow it to occur in the presence of a linear, scattering, intermittent video signal. By using an analog filter as part of the detection and data decoding method, the signal generated in the optical sensor is directly compared to a target frequency, which allows for fast and accurate frequency equalization and subsequent determination of the location of the data grid attached to the substrate. The optical sensor output analog electrical signal representing the label with the encoded information is subsequently quantized and decoded. By using an analog bandpass filter, label recognition can be performed without decoding the encoded label information. Once the center of the recognition target is found, the data grid report point can be determined. If the center of the recognition target is located in the center of the label, it is possible to identify the centers of the recognition target and the data grid simultaneously. The central placement of the recognition target on the label is preferred but is not necessary in the present invention.

An optically readable data grid of the present invention may encode 100 or even hundreds and more error-protected, alphanumeric characters spaced around an area of ​​approximately one square inch by encoding hexagons with three reflection characteristics in black, white, and gray. For sensors with an assigned optical resolution, the invention system allows for denser information compression than was possible with a bar code system. For example, if the high-resolution optical sensor is used in the system of the present invention, hundreds of alphanumeric characters may be encoded per square inch. Alternatively, as much as 100 wheels per inch can easily be used with the present invention in a transducer with relatively low resolution.

Optically Readable Labels of the Invention According to the present invention, optically readable labels may be produced with variable data densities due to the use of two or more contrasting optical characteristics. At high data densities and introduction of the recognition application into the system of the present invention, it would be desirable to have a more sophisticated spreading device with additional, more laborious, decoding algorithms required for reading encoded messages as compared to a reading system using barcodes. According to the present invention, data encoding may be performed by encoding a set of binary bit sequence into a beam of adjacent hexagons, each of the hexagons having at least one of two optical characteristics, although the encoding could be performed sequentially from one hexagon to another. A digital bitstream may be generated on a computer from data entered manually or otherwise converted to a binary bitstream, or may be pre-recorded as a digital bitstream. The data to be encoded is decompressed by a set of bits encoded in a given geographic area of ​​the data grid to increase the number of transitions between hexagons having different optical characteristics.

In a preferred embodiment of the present invention, the messages to be encoded are divided into high and low priority messages, which are arranged separately in different geographical areas of the data grid. High-priority messages can freely repeat in low-priority areas to reduce the chance of losing high-priority messages due to spreading inaccuracies caused by dirt, cracks, bends, and other types of damage to the data grid. High-priority messages are encoded in the central part of the data grid, adjacent to the recognition application in the preferred embodiment, to protect messages that are more likely to occur in peripheral areas of the data grid. The ability to correct inaccuracies is desirably entered into a data grid utilizing the large information capacity provided by the present invention, guaranteeing a high degree of data retrieval when decoding messages. The invention utilizes a mesh with high density pixels for printing labels with hexagons of various characteristics, although other printing techniques may be used without departing from the spirit of the invention. The grid with pixels is presented in such a way that after the label is printed, the optical characteristics of each hexagon are predefined, so that they can be later decoded to recover the data that was asked by encoding the individual hexagons. This type of printing process is well known in the art and standard printing devices and bit mapping techniques can be used to print hexagons having the optical characteristics necessary for the present invention.

In accordance with the present invention, a new and improved method of recovering data encoded by polygons, in particular hexagons forming a data grid, in a grid-encoded bit is created. The encoded labels can be passed through the encoded illuminated area and read optically by an electronic circuitry controlled by an optical sensor, or the dispensing device can be extended over the labels. An optical transducer produces an output signal, which is an analog electrical signal corresponding to the intensity of an individual region of the reflective power label recorded on a single optical element of the transducer. The optical sensor, by means of an analog filter, first aligns with the target magnitude frequency corresponding to a frequent detected target if it is on a data grid, once a good alignment is found, the label is recognized and the center of the target is detected. on the data grid. The analog signal is simultaneously quantized by an analog-to-digital converter on a continuous basis and memorized in the intermediate video memory. Recorded digital data representing the entire label is accepted for further processing in the decoding process.

The logic diagrams transform the digital data into a hexagonal interface card with different optical characteristics.

In the preferred embodiment of the invention, this is accomplished by calculating a standard deviation of the intensity of reflection characteristics recorded by an optical sensor on each pixel and assigned to a group of pixels surrounding this first pixel. As a result, large standard deviations correspond to areas of contrasting hexagonal transition boundaries.

Digital data conversions continue, including orientation orientation and hexagonal filtering programs. The main stages of this process are:

1) Non-linear filtering of the digital image.

2) Determining the orientation of the label by first locating the three image axes (see Figure 2) and determining which axis is parallel to both sides of the label.

3) Determining the center of each hexagon and setting the gray level at each center.

4) Changing gray levels to bitstream.

5) Free adjustment of this flow error.

6) Free bitrate conversion to the given character set.

It should be noted that while the process of the present invention is described for hexagons having two or more characteristics, the process, namely, optical image adjustment steps for label creasing, abrasion, and 1.1, can be used for other types of labels and other polygonal lattices.

Other objects and other uses of the invention will become apparent from the detailed description of the invention. However, it should be emphasized that the detailed description of the preferred embodiments of the invention is merely an illustration and should not be construed as limiting the modifications and variations within the scope of the invention as will be apparent to those skilled in the art.

FIG. 1 is a top plan view of a recognition target consisting of concentric circles as described in the present invention.

Fig ·. 2

FIG. 3

FIG. 4

FIG. 5

FIG. 6th

FIG. 7th

FIG. 9th

- Optically readable labels having adjacent hexagons for top-down encoding of data as described in the invention.

- An optically readable label having adjacent hexagons with three optical characteristics, encoding binary data and having a cognitive application as described in the present invention is viewed from above.

- A top plan view of a hexagonal three to three lattice array in which the hexagons are arranged adjacent and which can serve as the main encoding unit in the preferred embodiment of the invention.

- A field map graphically depicting a grid of data having 33 rows and 30 columns, which form a grid of 11 rows and 10 columns of hexagonal encoding block as three to three cells.

According to the invention, the control system cameras are designed to adjust the light of the optical sensor according to the height of the luggage, a schematic view.

- A detailed description of the decoding process according to the present invention.

- Block application that shows the process of locating the recognition target.

- Structure program and data stream for encoding and decoding process.

FIG. 10 - Step and Sequence Program of the Invention.

The ability to encode information by contrasting hexagons or lattices of contrasting colors arranged in a decimal pattern with a grid sequence and grid allows the information recorded on the label to be reproduced by means of an electro-optical sensor.

According to the present invention, the hexagons used on the label have some advantages in encoding the information on the label. These advantages are:

1. With a set optical resolution, hexagons can cluster more compactly than other polygons. For example, with the same resolution, the angles of the squares are difficult to distinguish, which in other cases does not require the resolution needed to read the squares. ' Circles would be the most optimal resolution, but the space between adjacent circles would be used aimlessly and complicate the processing of the label due to the need to give the spaces an optical characteristic. Hexagons allow for optimal information accumulation by comparing them with circles or other polygons consisting of octagons, squares, triangles and 1.1. Squares and triangles create problems because of their sharp angles. Circles and octagons create problems due to the resulting unused space between adjacent circles and octagon.

2. A grid of adjacent hexagons has three axes. The use of a square and hexagonal label may result in the principal axis of the hexagon appearing at a target ratio on the label side. This arrangement of the main axis of the hexagonal mesh makes it easier to read the data that would be enclosed within the hexagon in its ratio to this major axis.

In this case, the label comprises a discrete member with a corresponding adhesive-facing side which is attached to the article, to the outer surface of the container or to another article on which the optically readable information of the present invention is printed.

In this case, an optically readable data grid or data grid refers to an image of tangent hexagons or lattices having two or more optical characteristics for encoding in a rotary form, due to their respective characteristics and the spatial relationship between the hexagons. The hexagons printed with the content of this reproducible information are hereinafter referred to as hexagons with encoded information due to the way in which the information is encoded on the label.

This image is a hexagon of contacting hexagons with a maximum number of transition limits - a hexagon for optimal reading and maximum information memorization density, called honeycomb structure.

The contrasting reflection characteristics used to print individual hexagons or grid cells can vary greatly within the scope of the present invention. In this case, printing is the application on a substrate of material having the specified optical properties, or the change of optical properties when thermal printing is used. Printing is also understood to mean the omission of a material having a given optical characteristic applied to a part of the substrate when the substrate itself has a distinctive optical characteristic. For example, when printing a hexagonal grid in black or white, if the base is white, only the black grid should actually be pressed. Thus, in this case, the white hexagonal lattices are also term-printed or printed within the setting range.

In this case, the term optical characteristics include light absorption, reflection and / or refraction in different environments in printed lattices. When the cells are printed in black (high density black ink), gray (black grayscale) and white (absence of imprint on a white substrate) as in the embodiment of the invention, it is said that the invention has three optical characteristics.

In this case, as shown in FIG. 1, a set of concentric rings or concentric rings 10 means two or more concentric rings 12, one of which is the inner region of the 15th annular zone defined by the smallest radius of the rings r.

FIG. 2 shows a portion of an electro-optically readable label according to the present invention. As can be seen in FIG. 2, the label has a plurality of adjacent printed lattices as hexagons forming a honeycomb structure. Each of the individual hexagons is denoted by position 20 and has 6 smooth sides 22. The inner corners of the hexagon are also equal to each after 120. In the illustrated embodiment, the hexagon has a long vertical axis yy and a horizontal axis xx. The dimension of the hexagon by xx is slightly smaller than that of the hexagon by 20 by yy due to the correct geometry of the hexagon.

In the preferred embodiment of the invention shown in FIG. 3, which utilizes a label 30 having dimensions about once inches, which will be about 888 hexagons or lattices 20 (bearing in mind that in the preferred embodiment, the center of the label is occupied by a cognitive application 35 consisting of a plurality of concentric rings). These tangent hexagons 20 naturally form horizontal rows R defined by an imaginary line 31 and vertical columns C defined by an imaginary line 33. In this example, the label measuring one square inch has a total of 33 horizontal rows of R and 30 vertical columns of C hexagons 20. Each individual hexagon has a diameter of approximately 0.8 mm. In the square perimeter surrounding the cellular structure of the hexagons, due to the geometric arrangement of the hexagons converging, the rows R are larger than the columns C

Using the hexagons shown in FIG. 2, it can be noted that the hexagons are arranged in chess with overlapping vertical columns, and the alternating vertically arranged hexagons have collinear axes yy. The axes 20 of the arranged hexagons are yy in a linear interface with the outside vertical side 22 of the interlaced hexagon. The axes of the hexagons 20 are yy parallel to both vertical label boundaries 32 and 34 as shown in FIG. 3 Horizontal rows R are measured along the axis xx at the midpoint of the hexagon 20 .

As described below, the hexagons 20 are formed by a printing process in which the hexagons 20 are printed in two or more optical gradations, such as contrast colors. These colors may be white 25, black 26, also optional, but gray 27 is preferred, as shown in FIG. 3, although other contrasting colors may be used. Only two contrast types of white 25 and black 26 can be used, as shown in FIG. 2. The contrasting embodiment of the invention employs three contrasting colors: white 25, black 26 and gray 27 shown in FIG. 3. Specific white, black, and gray bars are selected for optimum contrast with electro-optical identification. The gray color is selected such that the optical gradation is approximately equal to the white and black colors used to create the label.

The label 30 shown in FIG. 3, may be obtained by the use of a discrete label having an area of ​​one square inch in the embodiment of the invention, or by using an acceptable color background (preferably white), the label may be printed directly on the packaging surface without producing a separate label. Since the presence of a background in the optical characteristic is essential and important, it is preferable to use a separate label for one of the contrasting colors, because the background color of the label is easier to control.

The alignment of the hexagons printed on one label with respect to the edges of the label is important for further determination of the major axis of the label, which is discussed below. The label is printed such that the honeycomb forming the hexagon axis yy becomes parallel to the vertical edges of labels 32 and 34 as shown in FIG. 3.

When reading a hexagonal grid, for the purpose of decoding the information contained in individual hexagons, it is important to have accurate color contrast between adjacent hexagons. Further, the less optical characteristics used to encode hexagons, the simpler the scattering device and the mathematical provisioning needed to decode the hexagons. But with fewer optical gradations, the density of data on the label also decreases. In the search for a compromise between the number of decoded information that can be stored on a label and the spreading price of labels with most optical characteristics of chaLT 3516 B, it was found desirable to print encoded hexagons with three optical characteristics, namely black, gray, and white. If the base or label has a good white background, when white hexagons can be filled with ink, you only need to actually print gray hexagons.

In the preferred embodiment of the invention, the hexagonal grid is created by printing the grid in black ink, but only every fifth grid pixel dot matrix printer in the illustrated example prints this way. This is accomplished by a semi-automation algorithm known in the art. This allows you to print with a predefined proportion of pixels that define a given gray hexagon, so when printing a black hexagon you need to print every single element of the image that determines that hexagon. A specific semiautonization algorithm used for printing labels according to the present invention is in the list of source code labeled Label, attached to microfiche, page 29, lines 39 to 48.

Black hexagonal lattices can be formed by printing them in standard black ink. As described below, the mathematical provisioning of the decomposition analysis decoding process allows a strong distinction between black, gray and white characteristics without the need for accurate color adjustment. On the other hand, if colors other than black, gray and white are used, or if different gradations of gray are used to create four- or five-color data grids, the contrast of the ink tones needs to be controlled much more closely to ensure the difference in optical performance between individual colors. It should be noted that the use of black ink is the simplest and easiest option for developing a cellular mesh with three optical properties of hexagonal lattices and is the most suitable embodiment of the present invention.

Due to the fact that, in the preferred embodiment, the label is square and due to the origin of the hexagonal lattice, the cellular structure edges have incomplete hexagons 56; as shown in FIG. 3, these incomplete hexagons are not used to convey any usable information.

In a preferred embodiment of the invention, the label also has a recognition application. The recognition target 35 shown in FIG. 3, made up of many concentric rings of contrasting colors (shown as black and white).

The black rings are marked 42, 46 and 48 respectively, and the white rings are marked 44, 50 and 52 respectively. The application is most conveniently placed at the geometric center of the label to minimize injury or completely or partially if the periphery is creased, soiled or damaged. Also, the size of the image buffer (described below), which is necessary to memorize data read from the label before the label itself is identified, is minimized when the recognition target is located at the center of the label.

The number of concentric rings used in the application may vary, but six concentric rings 42, 44, 46, 48, 50 and 52 have been found to be necessary and sufficient and their effective resolution ranges from white to black, etc.

Structure correlation technology is used to align the computational structure in which concentric rings are expected to be integral with the readable structure. When debugging occurs, the recognition target is searched as detailed below. A special filter designed and used in the preferred embodiment of the invention can be found in the appendix on microfiche, page 41, lines 5152, page 42, lines 1-8 and page 40, lines 19-41 with the heading Search C.

The recognition target can have any overall diameter smaller than the data grid, which can be 25% and most preferably equal to about 7% of the data grid space. In most cases, the recognition target is made as small as possible because the space it occupies on the label cannot be filled with coded information. Preferably, the diameters of the rings to be printed are selected such that the outer limit 52 of the outer ring 52 is about 7.45 mm. In this way, FIG. 3 area 35 of the recognition target represents about 7% of the surface area of ​​the label per square inch 30. Here, a satisfactory recognition target 35 may be printed on a label 30 per square inch without affecting the amount of information that may be encoded in the hexagonal grid surrounding the recognition application. In the case of incomplete hexagons on the outer periphery of the label 55, piecewise hexagons in contact with the outer boundary of the recognition target 56 are not used to encode the information. The width of each ring must be approximately the same as the dimension of the hexagons on the border / axis xx in fig. 1 /, this allows for improved resolution. Six rings fit. This is a valid number that allows the rings to be present on the minimum label area with the minimum number of possible misreadings due to incorrect markings on the label and other incorrect tags not on the label itself but on the conveyor belt.

The recognition target may take a shape different from concentric circles. For example, squares, spirals, or hexagons can be used to create transitions of contrasting concentric shapes, because linear transitions through the recognition application allow for regular, tasked, and identifiable color transitions that can be received by an electro-optical sensor and measured by a suitable filter. It should be noted that, although a helix and a set of non-concentric circles, a close approximation of the concentric circles can be obtained depending on the dimension and radius of the helix. Targeting from concentric circles is most appropriate because the frequency of the signal received through its center is the same as when the intersection in different directions through the center of concentric circles is performed. This helps to simplify the identification center, which will be discussed below, and enables the identification of the placement of the recognition target while simultaneously searching for an analog and digital broadcast signal, although the inventive method allows alternate or sequential use of high precision two-dimensional digital search.

In this case, concentric rings are called full rings, partial rings having a semicircular shape, sectors of concentric rings from 180 degrees to 360 degrees, and concentric spirals close to concentric rings.

Because each hexagon can be encoded with three different optical characteristics, the most suitable 1585-bit variant can be encoded in each hexagon (log. 3), employing more or less characteristics, the number of bits encoded in each hexagon will change accordingly. Coding algorithm

Obviously, if the three optical cha23s are designed to obtain data densities as close as possible to the maximum and increase the cell-to-cell optical chaperones, it facilitates the reproduction of the two-dimensional encoded characteristics.

FIG. 4 shows a set of cells 3x3 consisting of nine hexagonal cells 60, which is the basic coding unit used in the preferred embodiment of the invention. This is a sufficient condition in coding, but not essential. Other coding blocks are possible in the present invention. As shown below, 3x3 sets of hexagonal cells 60 are mapped to encode bit information 13 if the set is exactly 9 hexagons or less than 13 bits if the set is not full due to unused hexagons. 1292 bits of information can be written on a single square inch label with a data grid of about 888 hexagons and a recognition application occupying about 7% of the label area.

When encoding each set, the outer, lower hexagons 62 and 64 each set 60, as shown in FIG. 4, bounded by their respective characteristics, they are therefore determined to be distinct from the intermediate and tangent hexagon 66. Thus, one bit of the hexagon can be encoded in hexagons 62 and 64. This can be encoded in a set of 13 information bits 60 thanks to 11 bits in the remaining seven hexagons. . Due to the fact that 7 hexagons encode more possible combinations than used (for example, 3 hexagons)⁷ - 2187 combinations against 2¹¹ = 2048), some combinations are skewed, such as all black, all gray, all white, or essentially all black, gray, or white combinations. The need for contrast colors for the hexagons 62 and 64 relative to the hexagon 66 results from the guarantee of transitions required for the reproduced and free normalization tasks described below, as well as the need for use in determining the horizontal alignment of the data grid as described below. When encoding sets have 7 or 8 hexagons, the 7 useful hexagons are encoded by 11 bits, and the eighth hexagon, if any, encodes with one bit. For all other subassemblies, 3 bits are encoded on each pair of hexagons and 1 bit on each remaining single hexagon, as described below.

It is clear, therefore, that the label is highly efficient, easy to read (with the aid of an appropriate dissemination device and analytical mathematical provision), and a very high density information encoding on a relatively inexpensive, easy to print label. As noted above, the preferred embodiment uses a hexagonal layout of 33 rows x 30 columns per labeled area, with the cognitive application accounting for about 7% of the total label surface area. In practice, 13-bit information is derived from a set of 9 hexagons, so that 1.44-bit information is allocated to the cell. This is less than the theoretical 1,585-bit hexagon due to other algorithmic coding features, since not all 3 structures are used and some of the least optically desirable grid-to-grid transitions are removed.

From the following considerations, in a preferred embodiment of the invention, it is desirable to introduce a certain amount of error prevention in the encoding of the label so that the actual amount reproduced on the label during the decoding process is reduced to a high degree of data integrity.

It is to be noted that while the preferred embodiment of the label is shown and described, various variations of the label are possible without departing from the spirit of the invention. For example, the label does not have to be one square inch of 3516 B wheel inch. One square inch was chosen as the most appropriate, allowing for an acceptable data density of 100 alphanumeric information characters with a high degree of error protection without creating a large size label. It is desirable to have one square inch label to reduce paper and other costs due to printing, transportation and processing of these labels. Conventional labels with bar code for such measurements had a lower density. The use of 4.5 or more optical characteristics or color rendering can pack much more information into the most appropriate hexagonal space required, and the mathematical provisioning and sensitivity of the dissemination system to reproduce this information will become more complex. Thus, for practical reasons, a coding system having three optical characteristics: black, gray and white is highly desirable. Also, the dimensions of the hexagons and the recognition target can vary widely within the scope of the present invention. Although the division of hexagons into 3x3 cell sets is described above, other set structures or packages may not be used at all, and coding algorithms may be assigned to a specific individual hexagon structure. The relative number of coded information consecrated to the message, to the contrary, may also vary widely within the scope of the present invention.

The coding process of the present invention described below is to be used for the most appropriate label variant. It is to be understood that the preferred embodiment is disclosed below, but numerous variations, combinations, and rearrangements are possible within the scope of the present invention.

The process may begin with the most appropriate sequence of data to be coded on the label. In a preferred embodiment, the label is weighted and the data is divided into two fields, encoded as a high priority message and a low priority message. But, it is necessary to clarify that the invention is not limited to two messages or priority levels. Most message and priority levels can be created based on quantitative restrictions on the amount of labels and boxes in a given dimension.

For example, when a label is to be weighted, high-priority messages can take up to nine characters to represent the postal code of the recipient's cargo, parcel, or letter. Nine has nine numbers because while most individuals and companies have five-digit zip codes, nine-digit zip codes are becoming more common. Therefore, the postal code is a very important part of the information when handling a consignment. It determines the main purpose of the cargo and allows the use of various dispersion systems and control of the cargo so that it moves in the right direction on the car, airplane or conveyor system, etc.

Low priority messages can have, for example, a name and a loading address, including the postal code of the consignee, as well as billing information.

The reason for creating high-priority messages and low-priority messages is the need to protect the high-priority message from correcting the error, allowing you to place (encode) a high-priority message in the central part of the label that is less likely to be damaged or destroyed repeating and sorting the high priority message in the low priority message so that if even a high priority message is partially destroyed, there is a good chance that the high priority message could recover from the low priority message. Because of the high priority message being placed in the center, it may only be necessary to decode the high priority message for some purposes, so that only part of the label will need to be processed, which will allow for faster processing. This can happen, for example, when the cargo is on a conveyor and you only need to set a zip code to find out which of the several conveyor paths the cargo has to pass through.

Due to low priority, the low priority message is not displayed twice on the label. But, as described below, both low and high priority messages may have different codes with error protection and correction device to maximize the probability of both messages being accurately reset.

The use of the error protection symbols as part of the encoded information in the most preferred embodiment of the invention may be combined with a correspondingly written program and computer. forcing the system to correct errors during the decoding process as described below. The use of error correction is well known in the art and is available to those skilled in the art.

In the practice of the invention, the label operator can manually input data into an appropriate computer terminal for connecting, as shown below, a printer printing a label with a high priority message and a low priority message encoded in hexagonal labels, respectively. It does not matter if the high priority message and the low priority message will actually be created, but it is desirable to allow the key data for coding to be reproduced. In a preferred embodiment of the invention, the label is also printed with a centered recognition application having the majority of concentric rings in two alternating contrast colors, in addition the colors are best selected for two hexagons and most suitable for black and white for maximum contrast.

The operator manually entering this data does so that a properly programmed computer encodes each outgoing message character using a field setter to create a binary bit sequence of message characters encoded in the managed computer in a high priority message and a low priority message from each relative position. for unmasking. This operation is performed by the TEXTIN.C program, which is in the appendix, microfiche, page 1, lines 8-54, page 2, lines 1-54 and page 3, lines 1-36 and marked with position 110 in FIG. 9. A Compaq Deskpro 366 computer (with a 16MHz task frequency and an INTEL 80387 coprocessor chip) can be used.

The process may start with the information that is already encoded in the binary bit sequence provided, for example because it was received from a memory device or otherwise created. Therefore, the message to be encoded can exist in a form that is manually (electronically) converted to a binary bit sequence or that begins as a binary bit sequence.

As soon as the binary bit sequence is created or error-protected bit sequence at the stage discussed below, the bit sequence must be mapped according to the decisive mapping structure for encoding the hexagonal cellular structure of the present invention. FIG. Figure 5 shows a set card showing individual hexagonal cells in sets of 3x3 cells aligned into a grid or cellular structure containing 33 rows and 30 columns of hexagons. Each row is numbered and each column is numbered. Sequence numbers range from 1 to 33 and column numbers range from 1 to 30. It may be noted that some hexagons marked along the upper surface of the card and the right surface of the grid in the geometric center are denoted by x. This indicates that these hexagons do not have bit-coded information. This occurs because externally the x's are partial hexagons on the edge of the label, resulting in one hexagon less in each of these rows. Inner hexagons denoted by x form spaces either occupied by the recognition application or incomplete hexagons around the perimeter of the recognition target such that these inner hexagons marked with x are not mapped to bits. All hexagons that are not marked with x are able to write down information. In a preferred embodiment, each of these spaces should be occupied by hexagons in black / B /, white / W /, or gray / C /. As noted above, although various wrapping and mapping techniques may be used, the use of the present invention permits the use of packets for encoding specific information bits of 9 hexagons in three rows and 3 hexagons each, and as already described, preferably 13 information bits in each such packet. 9 hexagons each.

A data grid having 33 rows and 30 columns of intersecting hexagons is created and can be considered in relation to FIG. 5, a grid of 11 rows and 10 column hexagon sets, each with a converging hexagon structure of 3x3 lattice. It should be noted that each set of 3 rows of cells on a 3-column cell within a grid of 11 sets x 10 sets has a set of 7 or 8 hexagons due to the geometric placement of the hexagons and the number will change from row to row. In this way, 6 sets of 8 hexagons and 5 sets of 7 hexagons are obtained at this arrangement. Also, the centered recognition application allows you to create additional incomplete sets. In this way, FIG. 5 shows a graphical representation of the sets used in the hexagons. The hexagons here encode bitwise information in a data grid like 33 rows of 30 triangles overlapping columns.

As shown in FIG. 4, sets of nine hexagons are encoded by the following algorithm:

Take the eleven bits of information and map them in a sequence of seven hexagons marked a, c, d, e, f, and h.

Hexagons g and i use bit I to represent each so as to guarantee that each one differs from hexagonal h.

In this way, thirteen bits of information encode a complete set of 3x3 cells of nine converging hexagons.

For partial sets of 7 or 8 usable hexagons:

Eleven bits of information are retrieved and sequenced in the sequence of the first seven hexagons used. The eighth hexagon, if any, is used to represent a single bit.

For all other partial cells:

Map three bits of information into as many pairs of hexagons as possible. Uses any remaining unit hexagons to replace one bit.

Because seven hexagon mapping gives you more combinations than eleven bits (ie⁷ = 2187 before 2¹¹ = 2048), some hexagon combinations must be discarded. Rejected combinations become the ones with the lowest number of transitions. In order to accomplish this, reference tables are created for mapping sets according to FIG. 5. The creation and use of these reference tables can be done by a qualified software engineer. Accordingly, FIG. 9, the help table generator ΒΙΝΗΕΧ, LUT 132 and HEXBINLUT 134 can be found in the microfiche appendix, page 4, lines 3-52, page 5, lines 1-53 and page 6, lines 1-34 and labeled MKHEXLUT 130.

The use of this bit layout scheme allows encoding on a 1292-bit information data grid of 33 rows of x 30 columns of converging hexagons.

The sequence in which the high priority information and the low priority information are spread throughout the sets card is most appropriate depending on:

a) High priority message size,

b) Low priority message size,

(c) Optimal placement of the high priority message in a protected location.

Using the kit card, FIG. 5, as a program template for MKMAPS, the C 140 mapping memo, working with digital data contained in the memorization environment, performs the primary action of distributing information: Throughout the set card and high priority messages and low priority messages, this will be described below. The coding program is shown in the accompanying code list MKMAPS.C 140 and can be found in the microfiche appendix on page 19, lines 3-53;

page 20, lines 1-53, page 21, lines 1-53, and page 22, lines 1-42.

In order to reduce the possibility of error and to ensure the possibility of error correction, the preferred embodiment of the invention must have extensive error protection and correction capabilities. For example, in a preferred embodiment having 1292 bits of information that can be encoded in a hexagonal grid of one square inch, we have 33 rows x 30 columns of hexagons, and a recognition application occupying about 7% of the label, preferably using 36 bits of information to encode the high priority message. 9 - postal code, plus one additional alphanumeric character, which must represent the loading code. This example should also use 120 check boxes indicating the high priority message. This is determined by the number of corrections to the required opportunity error. In the example shown in the example, the 560-bit low-priority message, that is, the 40-bit high-priority message that is turned into the low-priority message. In the example shown, 516 low check bits are added to provide security and facilitate the recovery of low priority message. This example illustrates a more grateful use of check bits to protect and have the ability to restore high priority messages as opposed to low priority messages. It is to be clarified that the information provided is merely an example and the high priority message may have been longer or shorter, the low priority message longer or shorter, the number of control bits higher or lower depending on the correct use of the invention.

priority message to create

The system code takes on a specific message sequence and adds a different error checking sequence and message sequence. The non-systemic code takes the sequence of the specific message and enters the error checking sequence with the message sequence so that the message is no different - but is, of course, reproducible. The use of systematic or non-systemic coding for error protection is within the scope of the present invention. The following is the system code.

It has already been shown that the step of inserting error-detection symbols consists of systematic and / or non-systematic coding.

Various systematic linear cyclic codes for protection against error are known in the art, such as BCH codes, Reed - Solomon codes, Heming codes. In a preferred embodiment of the invention, the Rido-Solomon codes are separately enabled for storing high and low priority messages. Rido - Solomon codes are very effective and most useful when checking for multi-character errors. Rido-Solomon codes are well known and need to be explained that this is the preferred option, although other codes with error correction may be used in the invention. Rido - Solomon and other coding systems are described, for example, in Theory and Practice of Error Control Codes, by Richard E. Blachut and Edison Wales, 1983. pages 174-175.

The following are examples of information about Reed Solomon's Code. Specific characteristics of the Rido-Solomon code:

m = number of bits in each character, n = number of characters in a block - 2⁶-l, k = number of message characters, (number of message bits = km), t '= correction power in number of characters = (nk): 2

The multi-digit zip code and single-digit alphanumeric character require 36 bits without error protection for further identification purposes, as described below. The Rido-Solomon code with these parameters was selected for the high priority message.

m = 66 - bit characters, n = 2^e-l = 63 t = 10 so k = n-2t = 43

Due to the fact that only six 6-bees are needed for the 36-bit message, the remaining characters / 43-6 / are redundant characters that are meant between encoders and decoders, and do not need to be labeled. This way, the total number of bits required by the label for high priority message is / 63-37 / x6 or 156 bits.

This encoding scheme will be able to correct errors up to a maximum of 60 (10x6) bits, representing up to 38.5% of the bits used. Due to the large number of available redundant characters, the high ability of Reed-Solomon coding to detect errors makes it absolutely impossible to read high-priority messages incorrectly.

The low priority message was encoded in a Rido-Solomon code with error protection with other parameters, namely m = 8/8 - bit characters, n = 2⁶-l = 255 t = 36 x = n-2t - 183

Because there are 1292 bits on the label for encoding according to this example, the total number of bits 1336 bits (1292 - 156 bits for high priority message and check bits) is for the encoding and check bits of low priority message. That way, the remaining 904 bits / 255x8-1136 / are to be understood as excess bits. This allows for 560 bits / 183x8904 / for informational content of low priority and 576 check bits.

For a more reliable restoration of the high priority message, it is also in the low priority message. Error-protected Reed-Solomon code used for low-priority messaging allows for additional 866-bit alphanumeric encoding and has a maximum error correction power of about 25.4%.

Using the Riddle-Solomon error-prone code can provide general information from 1292 bits on an illustrative label, as follows:

high priority information bits

120 check priority bits

560 low priority information bits (embedded in low priority messages)

576 low priority check bits

A sequence of data bits combining respective information save check bits for individual hexagons in the assembly card of FIG. 5.

It should be noted that there are many different distribution schemes that can be used, bearing in mind that important criteria to determine are:

1) Safe positioning of high priority message near recognition target (if displayed on data grid)

2) Creating a structure within reasonable limits for repeated assembly reading.

The specific error coding program used in the example is in the microfiche appendix titled EPPCODE.C on page 15, lines 1-52 and on page 16, lines 1-50.

Coding with Rido-Solomon codes requires Perform vector code message multiplication from a generator matrix. Matrix multiplication is performed using Galua field arithmetic. The addition of any two field elements is obtained by performing or action between component elements. Multiplication is obtained by logarithmic field Galua field. Get logarithms and antilogarithms using reference tables consisting of the first polynomials for extremely high priority messages: l + x⁶; and low priority messages: l + x²+ x³+ x⁴+ x⁵. As shown in FIG. 9, the GF.C 126 utility provides compilation tables for Galua field arithmetic. The GF.C utility can be found in the microfiche appendix on page 8, lines 1-53, and page 9, lines 1-32. Help tables are created and stored in GF.LUT / 127 for use during encoding and decoding. The generic polynomial q / x / for the Rido-Solomon code is determined by the following equation:

q / x / = / x + a // x + a²/. . ./xa^2t/ where a base element of the Galua field.

The generator matrix for the Rido-Solomon code is formed by the long division of each row of the generator matrix. Each row of the generator matrix is ​​given the residue that occurs after the long division by dividing by x^n_k_1of q / x /.

Calculations of the generic polynomial q / x / as well as the geometries of the high and low priority messages are performed by the utility MKRSLV.C 125, which is in the microfiche appendix on page 10, lines 1-52, page 11 lines 1-53, page 12, lines 1- 54, page 13, lines 1-52, page 14, lines 1-4. Reference tables of generator matrices are created and stored in file RS.LUT 128.

In a preferred embodiment of the invention, the labels containing the hexagons are printed with standard equipment, which is always available and inexpensive. The printing device, which has a 300x300 dot matrix signage, produces good results when printing tri-color (black, white, gray) labels with 888 hexagons and a centered recognition application. This printing device is compatible with Hjulet-Pakkard's Laser Jet 11 Series and has a 0.5 megabyte of memory and 300 dots per inch graphics resolution. The 300x300 pixel grid, which has a density of 90000 pixels per square inch, can produce about 90 pixels into a hexagon in a preferred embodiment. Each pixel is assigned a value of 0 or 1, which corresponds to a black or white pixel. This printing device is used to print a two-color grid of black or white hexagons. It can also be used to print a tri-color grid of black, white, and gray hexagons, provided that the grayscale algorithm for creating gray hexagons is used, as described above.

As depicted in FIG. 9, a help table PEGIONS.LUT 141 was created by means of the program MKMAPC.C 140 from 34 rows and 30 columns, which is analogous to that shown in FIG. 5, but which is only adapted for black or white selection markings, recognition ring. Individual hexagons are coded as black, white, or gray, or as unused. Separate help table ΗΕΧ MAP.LUT 142 is created by a written subroutine of the MKMAPS.C program, which queries the dependency of each of the 300x300 pixels on a pixel grid for specific areas of REGIONS.LUT 141, which is about 90 pixels on hexagons. The pixels in the orientation rings are coded in either black or white. Recognition target rings for printing, first as a result of generating a hexagonal structure in each row and then generating rings. Areas partially or completely covered with search rings are considered as unused REGIONS. LUT 141. Previous MKMAPS.C program and subroutines can be found in the attached source list in the microfiche appendix, pages 19-22.

The encoded bit sequence, which protects against errors, is mapped according to the most suitable sequence into a prefabricated grid of 11x10 hexagons. As shown in FIG. 9, the sequence is queried for the ordered departmental table ORDER.LUT 151, generated by an auxiliary written program named ORDER.C 150, which is in the microfiche appendix, page 26, lines 1-47, and page 27, lines 1-3. The recorded program PRLABEL.C 160 on the microfiche appendix, on page 17, lines 1-54 and r8, lines 1-39, is used to assign values ​​0, 1 and 2 to areas in print, and the area with value 3 is left unchanged. The gray levels for each of the hexagons in the set of 3x3 cells are asked in conjunction with the written program CELL CODE.C 170, which can be found in the microfiche appendix on page 23, lines 1-53, page 24, lines 1-53, page 25, lines 1-43.

Priority is given to the high priority message memorization area that comes closer to the memorization target, where it will be less affected by label degrading, settles into this utility. As a result, the LABEL.C 180 program is used to create bits suitable for input to a laser print device. LABEL.C 180 can be found in the microfiche supplement, page 28, lines 1-53, page 25, lines 1-52, page 30, lines 1-36.

It should be noted that the use of black, gray, and white allows for a simple label printing procedure, all that is required is black ink, using a standard semi-automation algorithm well known in the art. If other color combinations are used (which is possible), the need for printing in other colors poses a fundamental problem compared to the tricolor black-gray-white version or the two-color black-white version.

In this way, when each of the printer pixels is given a black or white size, the labels can be printed to create an encoded format as shown in FIG. 3, where some hexagons are white, some are gray and some are black, and where the recognition target area - the most appropriate black and white concentric rings - is in the geometric center of the label.

After describing how the data is written on the label and how it is printed, it is also necessary to describe the interpretation of the label or the decoding process. It should be noted that it is desirable to perform the interpretation function at high speed, in a fraction of a second, in order to increase the efficiency of cargo handling (or other processing or label reading).

There are two main paths that can be used in the process of reading a label to receive an image. The label can be read at relatively low speeds using a manual statistical dispenser with a fixed focal point. Alternatively, it is very convenient to use an electro-optical sensor with a servo-controlled focusing mechanism, which allows the dynamic dispersion of rapidly moving variable dimensions and high payloads to be achieved very quickly. The decoding process and the device described below have been demonstrated by the use of a fixed focal unit. The process having the basic capabilities described herein with respect to a static scattering device with a fixed focal point can be applied to a dynamic scattering system with defined optical system modifications, as described below. For high-speed handling, it is desirable to have a high-speed dispersion mechanism that allows labels to be read at a linear speed of 2.5 m. per second or more and passing under a fixed reading device. In this way, the image processing function consists of the following steps. FIG. Figure 7 is a schematic diagram of the steps of the decoding process.

1. Label lighting

When a packing, parcel or letter passes through a high-speed conveyor, the area for illumination appears large enough because the dimensions of the conveyor-adapted packaging can be large and variable. For example, cargo handling systems have an unusual lm conveyor and packaging ranging from a few inches to a meter (and similar heights). Therefore, a one-square-inch label can be placed anywhere on the conveyor. It is possible that the packages will also be angled with respect to the axis of movement of the conveyor belt. Packages, Packaging, Letters, or similar items may be of various heights such that scattering labels may appear, for example, over the conveyor one inch or less on one side, or up to 90 cm or more in the maximum amount of Packaging that can be processed by the system described. in height, on the other.

For the purpose of illuminating labels in accordance with the present invention, particularly with respect to a wide range of package widths, heights and labels, it is desirable to utilize a high-intensity light source that well reflects the optical characteristics of or the visible and used visible light spectrum may vary. Light measurement technique usually consists of measuring light reflected from a black, white, or gray labeled hexagon.

The light source shall produce sufficient light reflected to the light transducer (for example, a device with a charge connection as described below) so that the light transducer can reliably distinguish the optical characteristics of the hexagons used for measurement between black, gray and white or otherwise. The dynamic scattering system may utilize an LED grid to create an illumination level of 10 bmt / cm in the illumination area of ​​the label at the label level. The LEDs can be located in a bulk grid without focusing lenses or in a linear grid using cylindrical focusing lenses. The laser light source transmitted through the respective optical system constituting the linear illumination source may be used in the present invention.

Selection of light source and source properties for the application being considered within the skill of the art. It is important to remember that the label you are looking for can only have a maximum of one square inch at a maximum height of 90 cm on a 1m wide band moving at a speed of 2.5m / s. It is very important to illuminate the labels so that those labels are well identified and discovered.

In the case of a static transducer with a fixed focus, which is used in this example, the illumination level of 2 MVT / cm proved to be sufficient for the practice of the invention. This was done by a luminescent light source.

2. Optical Measurement of Reflected Label Image

The second step in the decoding part of the tuning part is the optical measurement of the illuminated part by an electrically controlled sensor. The hardware light sensor used in this example for a static focal point broadcast system incorporates an industrial quality color television camera, charge model WV-CD 130, from Panasonic Industrial Company, Van Panasonic Vej, Sekokus, New Jersey 07094 TV lens with 5mm extension tube, 50mm focal length, 1.3 luminous intensity, DO Industries, Ink (Japan), 317 Ist Chestnat Strit, Ist Rochester, New York 1445 and the NAVITRONT trademark. The camera is connected to a coverage plate of the invention designated DT-2803-60 by Deita transleišen Ink, 100 Lok Draiv, Malboro, Masajiets 01752.

Optical measurement can include an overview of the entire label, using the above-described camera-type spatial sensor and the scope plate of the invention, or alternatively, a linear grid sensor having a device with a chip chip in which a second measurement of label dispersion is performed by packaging (and label). movement. For this purpose, the Chip Tomson CSFTNX 31510 SD3 Charger Unit, a 4096 Charge Link High Speed ​​Linear Sensor provided by Tompson CSF, Divižin Tjūbs Elektronics, Clica Voter BP 305 92102 Bulon-Bijanjur Seating, France is suitable.

In dynamic systems where the labeled cargo moves on a conveyor, it is desirable to have a long optical path between the recognizable labels and the light source. The first purpose of creating a long optical path is to reduce or increase the apparent change in dimension of the label as measured by a remote light sensor. For example, if the optical path is, for example, 1 m by 20 cm, the size of the visible label on the conveyor belt will be very different from the size of the labels at a distance of 90 cm above the conveyor belt. If an optical path of six meters is used, the apparent dimensions of the same labels will be the same. Thanks to this, it is possible to fill the whole or practically the whole area of ​​the light sensor with a measuring area, regardless of the height, which allows achieving a high resolution of the image. If a spatial sensor is used instead of a linear sensor, the same principle applies. This can be accomplished by the long optical path shown in FIG.

A height sensor is required to focus on labels of varying heights. An ultrasonic transducer may be used or the beam sequence may be interrupted as a transducer. Any of these systems may be utilized and may actuate a suitable adjustable focusing mechanism with closed or open loops for measuring and adjusting the position of optical measuring elements (such as lenses or transducer) depending solely on the continuous base shown in FIG.

FIG. Figure 6 is a schematic view of the focusing system and adjusting the position of the camera sensor according to the invention according to the height of the load being checked. Fig. 6 shows a lens 196, a coil drive, a height sensor and a feedback loop suitable for the invention. The height sensor of Fig. 206 may be an ultrasonic height sensor, or a beam of light interrupted by each load moving on the conveyor. The altimeter output signal is fed to a microprocessor element 204 that engages a coil actuator 202 that moves a coil 200 on which is mounted a device 198 with a spindle position connection or other suitable light sensor. The sensor 208 measures the position of the coil 200 and its output to the microprocessor element 204 completes the measurement and adjustment of the position of the feedback coil 200.

The transducer shall measure the reflected light reflected from the illuminated label and shall also produce an analogue signal corresponding to the brightness of the characteristics of the reflection label recorded on the individual pixels of the electro-optical transducer.

A suitable light source, as described above, may be placed on a conveyor on a stationary surface and include an area extending across the entire width of the conveyor with light of optimum brightness. The light reflected from the label may bend according to the reflector sequence and then be measured by an electro-optical sensor.

The purpose of a curved optical path is to create compact, and therefore solid, systems.

It filters the video signal from the output analog sensor. An analog electrical signal is used in conjunction with an analog bandpass filter to locate a recognition application on a data grid. The analog signal is then inverted into a digital signal using a simple analog-to-digital converter embedded in the image capture board described below or other known means. Instead of an analog bandpass filter, a digital filtering scheme can be constructed to identify the recognition target and compare the digital data representing it to the quantized output signal of the analog to digital converter, which will be described below.

An example of a spatial transducer having a chip-loading device with a plurality of detectors used according to the invention is the above-described color television camera on a discharge-type panasonic WV CB 130. The transducer output analog signal was fed onto the above-described streaming video frame. board type DT-280 3-60 having - bit monochromatic analog-to-digital video conversion for quantization and further processing. The recorded output of a suitable subroutine, arranged with the aid of a suitable subroutine, was stored on the memory device as an exact replica of the image recorded by an optical sensor.

3. Reflected image processing

An essential part of the invention is the processing of an optically measured image to re-create and orient the configuration and color (optical characteristics) of the primary label to each hexagon. This is done using the following steps, after which the known structure by which the label was originally encoded and bit-encoded can be used to decode the information contained in the label.

a) Finding the center of the target

Prior to use, the above described camcorder with charge communication device and image volume card shown in Fig. 10 were skipped by the DTWIT.C 250 minimization program for positioning the image volume card into a known ready position and loading the output department tables, followed by DTLTVEC.C 255. , which inserts the image volume board in live mode.

The DTGRAB.C program then controls the work of the image-volume board based on the quantization of vision in the image memory into 250 rows and 256 columns, memorizing the samples as 6-bit sizes aligned right bytes. The mentioned programs can be found in the microfiche appendix in column 31, lines 1-53, page 32, lines 1-39, page 33, lines 1-22 and 34, lines 119. Auxiliary programs DTSAVE.C and DTLOAD.C allow to transfer the screen image to in and out of the storage device. Source code lists for the given programs can be found in the microfiche supplement on page 35, lines 12-33, 36, lines 13-33, respectively.

A standard analogue bandpass filter may be used to initialize the label image to identify the characteristics of two or more concentric rings of the recognition target. Both characteristics are best represented by black and white because the highest energy signal is the highest contrast. To find a fixed transition structure from black to white and again to black and so on it is desirable that linear scattering by the recognition application and passage through the center of the target produces a uniform frequency response regardless of the label orientation. In this way, the target rings are optimally composed of contrasting concentric rings. The output of the transducer is then doubled and selected in two directions of detection. In one direction, all the energy in the output signal is detected, and the other energy is measured in the ring frequency. When comparing the output signals, the energy in the ring detector is closest to the energy in the entire energy detector when the tile is measured along the center of the recognition target. The center of the recognition target is found at this near approximation, the source code lists attributed to the creation of the digital tape filter and the filtering process can be found in the microfiche appendix, file name IEND.C, pages 39-43. But in the dynamics of the preferred embodiment of the invention in the first filtration step, an analog bandpass filter or, in addition, a selective bandpass filter, although a digital filter, is preferred.

It is to be noted that the step of searching for recognition applications is designated FIND.C 280 in FIG. 10, shown as optional in FIG. 7, because the manual dispenser may be used in the method of the invention, in which case the operator may accurately position the dispenser to correct sensor alignment is guaranteed. This is obviously much slower than using an automatic sensor, and using an automatic sensor is best if you are working at high speed. If the automatic sensor (rather than the manual) is used at work, finding the target is a necessary step in the process.

As an alternative to the analog filter described above, a digital bandpass filter can be constructed using the Parks-Maclenon algorithm provided with the mathematical provisioning of the digital filter constructions on the IBM PC

Teilo · and Stauraitis, Marsel Dekter, Ink., New York, 1987.

One-dimensional digital bandpass filter is used in connection with the invention for filtering a normalized digital bit sequence, which is described below with the aid of the following filtering subroutine. The band to be filtered is the expected frequency of the ring. The one-dimensional digital bandpass filter was designed for a sampling rate of 400 pixels per inch and a length of 125 pixels (or 0.3125 inches), and was designed to work on the basis of the printed ring recognition target dimension as shown in Figure 3. Frequency 300/16 linear pairs per inch, with normalized frequency output (where 400 pairs per inch = 1 / B) 300/16 x 400, or 0.046875.The filter with a band 5% below this frequency and 15% above was selected because label distortions usually result in image shrinkage and therefore increase in frequency. Retention bands were made from 15% below frequency down to 0 and 25% above ring frequency up to 0.5 (Newquist limit). The filter coefficients were stored in the file IMPULSE.LUT 275, fig.10, for further operations the first 62 coefficients are excluded because the filter is digital. Fig. 8 shows a block diagram of the program. Further references can be made to the source code lists included in the microfiche attachments, filename FIND.C 280, starting on page 39.

A filter of 25 pixels of its own length was made due to the selection of a bandpass filter at output intervals corresponding to the measured horizontal gain. For example, if horizontal image gain is 80 pixels per inch, every * fifth filter selection will be used (400/80 = 5 pixels). For incomplete stages, linear interpolation of adjacent filter selections is used.

A second two-dimensional filter 25 on 25 pixels was also used. The sample sizes for this two-dimensional filter are based on the Euclidean distance of each point from the center of the filter, which has been scaled for horizontal and vertical gain, respectively. It then uses linear interpolation for incomplete sampling intervals.

The output of the specified one-dimensional filter was squared and then compared to a recursive low-frequency first-order filter providing an exponential, expired process window. When the output of the smoothing filter exceeds a specified threshold, it is not necessary to use a two-dimensional filtering step to confirm the existence of the target and determine its exact location, this is discussed below. In the first part of the two-dimensional filtering, I used a small-size 10-pixel-to-10-pixel filter, because of the computational economy. This filter emits a rectangular area around a layout defined by a one-dimensional filter. If the maximum two-dimensional correlation exceeds a given threshold, then the final stage of two-dimensional filtering with full 25 x 25 pixels was used for a small square window around the maximum. if the best result of this filter exceeds the set limit, the center is detected. If none of the limits were exceeded, the program temporarily unloaded the smoothing filter and returned to one-dimensional dispersion. If the one-dimensional propagation ended without the detection of the existence of the recognition target, the program exited with a return error. For any other filtering process used in the example illustrated, it is necessary to refer to the source code lists in the microfiche appendix, lines 39-42.

b) Normalization of the measured image

The brightness of the reflected light recorded by the optical sensor used in the process may vary due to variations in light, print density, paper reflectance, camera sensitivity, and other reasons including label damage such as bends, folds, and 1.1. As an optional but desirable step of reflection of light, measured by a sensor and transmitted to memory, it can be normalized by a standard procedure. Using state-of-the-art technology written in the normalization program WOPM.C. 270, shown in FIG. 10, was used to analyze the brightness levels of the reflected light from the label, recorded by pixel blocks in the scattering device, in order to find the maximum and minimum of the reflected light for the grid. The ordered digital output of the specified scanner and the combination with the image-size plate from memory were stored on a computer for further processing by a written normalization program.

Using the equation y = mx + b, where the minimum brightness placed at the x position would give y = 0 and the maximum brightness at the x position would give y = 63, the brightness values ​​recorded for each light-reflecting pixel were adjusted as follows: , so that the blackest color and whiter color in the captured image are maintained as standards, while other reflections of black, white and gray are raised to these standards. In this way, the normalization step facilitates the processing of the measured image. Normalization is performed using the recorded program NORM.C, which can be found in the microfiche supplement on page 37, lines 10-52 and 38, lines 1-11. It should be noted that perfect normalization procedures known in the art will be employed.

c) · Image computing

For subsequent calculations, the reflected image of the reflected label is recalculated to produce an image with uniform horizontal and vertical gain. Again, this is not a necessary step, but. it allows for quick and easy recovery. to make it vertical

To render the screening information encoded for the purpose of selection, the operation was performed in a horizontal and resolution-like manner, such as 150 pixels per inch, as shown in the static of the invention with a fixed focal point.

The conversion action occurs due to the selection of fine address rows and columns to 1/150 inch. based on known horizontal and vertical amplification deduction. Each pixel in the new solid-state converted image is then distinguished from its corresponding set of pixels in the replicated image. For point size approximations, small addresses use binary interpolation. Due to the conversion, the center of the label is placed in a known memory position. The converted image is memorized for further use in the search phase. Used in all subsequent steps of the process to re-center the label image! in a known position on the grid, but it should be noted that this does not indicate the orientation of the label, which may be asymmetric with respect to the transducer. The conversion operation is controlled by the recorded subroutine, which is listed in the microfiche on page 42, lines 14-52, and 43, lines 1-14.

(d) Two-dimensional synchronous resetting (sampling)

The following set of process steps is called two-dimensional synchronous reset. The steps are performed with the aid of a suitable program and subroutine called CIOCK.C 290, shown in FIG. 10 and located on pages 44-51 of the microfiche attachment. This operation is performed in two measurements on the converted image to accurately determine the position of the hexagon on the primary data grid. The purpose of synchronous restoration is to locate sampling points and correct defects in the crease, twist, or tilt label, because the label cannot be completely flat. This is an important part of the process and is not defined by hexagonal coding labels. This operation is also applicable to other encoded label decoding processes consisting of a regular two-dimensional grid and squares, triangles, etc.

One-dimensional synchronous reset is a common concept well understood in signal processing. Two-dimensional synchronous reset is a continuation of this process and will be understood by some skilled practitioner. It should be understood that the term synchronous reset is not misleading for the expert because it does not belong to synchronization.

(i) Edge reinforcement and non-linear machining

The first step of synchronous reconstruction can be accomplished by various non-linear mapping operations known in the art for generating signal components at a given synchronization frequency, which are output from an optical sensor in an image quantum output and an image volume board. The purpose of nonlinear mapping is to (preferably) capture the normalized and recalculated image that exists at this point during the process and form a two-dimensional linear card that enhances transitions between adjacent contrasting hexagons. In a preferred embodiment of the invention, this is done by standard error mapping. This step can also be performed by filtering the image differentiation branLT 3516 B, some filtering techniques are known in the art.

Laplace or Sobel type cores, and then determine the absolute size or square the results. These procedures can be found in Digital Image Processing, Raphael I, Gonsales, and Paul Vintz, Eds. Edison Weird, 1977

By plotting the standard deviation, the grid-to-grid image with undifferentiated edges is memorized. A standard deviation card is then created to determine the location of the tangent contrasting hexagons by defining standard deviations for the 3 x 3 sets of pixel groups (which are different from the 3 x 3 grid sets), for determining standard deviations of pixel brightness. Performs standard deviation calculations for local pixels with a fixed color (the smallest standard deviations that make up the center of a hexagon or the dividing line between monochrome hexagons, as opposed to groups of pixels with larger standard deviations that make the transition from one color hexagon to adjacent hexagon of contrasting color. Because adjacent hexagons often have the same color, the standard deviation card does not fully distinguish each hexagon. Due to the fact that the standard deviation mapping process does not distinguish the boundary between hexagons of the same color, these hexagons result in skipped boundaries or edges. Other aspects of the synchronous reproduction process are directed to the regeneration of these missed transitions.

In the preferred embodiment of the invention, the use of unnecessary technology can reduce the number of calculations required to form a standard deviation card. Calculating the sum of a nine-pixel image in a 3x3 pixel block would require octet composition operations. This can be reduced by doubling the sum of each element of the image itself by the amount of its own and the image elements to the right and left of it. This would require two composition steps for the pixel. The same operation is then performed on the new image, without the amount recalculated for the pixels directly above and below. This requires two more composition actions, making the total number of composition actions four. It can be shown that at the end of these steps each pixel is replaced by the sum of its immediate neighbors and its eighths.

Standard deviation mapping is a necessary technology in creating this hexagonal card, which corresponds to the primary data grid, but only with transitions skipping between the primary hexagons of the same color. The specific standard deviation mapping technology used for the example shown can be found in the source code lists in the microfiche appendix on page 45, lines 14-53, page 46, lines 1-4.

(ii) Framing

The following, called framing, is a subroutine that is optional. In the practice of the invention, framing was used to reduce the brightness of the boundaries that are not bound to the outline of the hexagon. These boundaries occur at two points: the target ring and the uncontrolled image that surrounds the label. The comparative function is used to reduce the brightness of these areas. The details of using framing as an additional tag for quicker Fourier replacement are in specialist copmetals. The framing procedure used can be found in the source code list in the microfiche appendix on page 46, lines 6-22.

iii) Two-dimensional fast Fourier replacement

The two-dimensional quick change of Fourier numerical values ​​corresponding to a (not necessarily) framed standard deviation card is then executed under the control of common recorded programs. During operation, the copier performs a Fourier quick change of the sharpness, transition direction, and transition limits of the image created in the previous stage for presentation of the two-dimensional layout, contrasting hexagons found in the standard deviation mapping step. In other words, a fast Fourier change is the border between hexagons when they are known for their arrangement, direction, and sharpness. In this way, the constant arrangement and directionality of the hexagonal boundaries will give the corresponding points in the exchange area a high energy level. The brightest point will be a point 0.0 in the exchange plane that corresponds to the DC and image components. The six dots surrounding the center point provide the arrangement, direction, and sharpness of the bezels between the hexagons.

Because the image is a real (and not a complex) size, the swap area is a point symmetric to the beginning of the coordinates. In this way, only half the plane of the swap area has to be calculated, which saves almost half of the machine time. Abandoning these calculations allows you to reduce the number of tests required for subsequent inversion of image filtering. Quickly used in connection with an illustrative example of a static system with a fixed fireplace, was available for R2D77T from software package 87 FFT-2 ​​by Maikrouei, Ink., Kongston, Massachusetts.

and the Fourier Fast Fourier conversion program,

4. Image filtering

The filtering process for the reconstruction of all the hexagons in the image area using paLT 3516 B strange digital data is still required. This can be done by turning off any points in the conversion area that do not conform to the specified layout and direction hexagons identified in the standard deviation mapping step. The six noticeable points in the exchange area are due to the hexagonal cellular label design. Only three points are realistically identified in the swap area, so the image is symmetric with respect to the beginning of the coordinates, and the second three points can be assumed from the first three. Most preferably, the filtering is done in three steps to avoid transitions that are too wide, too narrow, and / or misaligned, from the standard deviation mapping step.

First, it performs high-frequency filtering by zeroing all points within the circle around the beginning of the coordinates of the conversion area, but at some distance outward from the beginning of the coordinates, not enough for the six points in hexagonal graphical conversion. These dots are larger than spaces than hexagonal spaces and therefore provide information that is related to missed transitions in the label image. Removing missed transitions in the label image requires the removal of missed transitions in the Fourier Change area.

After that, all outside points of the radius in question are reset without the six noticeable points in the swap area. This corresponds to incorrect transitions, which are particularly close. This operation is combined with the first to form a ring of remaining points. Creating this ring is equivalent to creating a spatial bandpass filter. The inner and outer radii of the ring are determined by the apparent layout of the hexagon. Since the diameter of the hexagon as shown must be 5 pixels in the descriptive example, and the length of the conversion to 256 pixels will be 256/5 = 51.2 pixels from the center in the changing area of ​​the hexagon. It uses respectively a ring with an inner radius of 45 pixels and an outer ray of 80 pixels, corresponding to hexagonal diameters of 3.5 to 5.69 pixels. The most suitable filter for high-frequency skipping was used because the deformation of the label - creasing, skewing - increases the image's sedimentation.

After the spatial bandpass filtering described above, there is a ring with six conspicuous points, and each point has an even angular arrangement relative to the center (point 0.0) in the exchange area. Uses the filtering direction step to complete the task in the change area by eliminating unnecessary information. Any point at a very large angular distance from the visible areas in the changeover area is converted to zero. This leads to the removal of any edges in the image area that occur in one of three directions due to the hexagonal honeycomb oblique structure.

For directional filtering, it is necessary to find the most noticeable points remaining after the spatial bandpass filtering. This point is said to be one of the noticeable area points and substitution that resemble the vertices of a hexagon. Five other noticeable points within the same radius of the center and having a multiplier arrangement of 60 degrees are also evident in the exchange. As a result, all other points with an angular distance greater than 10 degrees from any of these points are eliminated.

Six edges of the ring remain. This step-by-step filtering process removes any information about the misalignment or direction of the image area. Removing this misplaced information allows you to restore a complete drawing of each hexagon in the change area.

The described filtering steps take place under the control of written subroutines that are in the source code lists in the microfiche on page 46, lines 26-52, 47, lines 1-52, page 48, lines 1-52, page 49, lines 1-46.

5. Fast reversible Fourier change

For true return to the image area, while restoring the drawing image of the interconnecting hexagonal data grids, a two-dimensional backward fast change of the filtered data filtering area Fourier (2D - IFFI) is desirable. Back-to-back exchange is done with the standard two-dimensional Fourier Backward Exchange (R2D1FT) program, which is available in package 87 of FF-2, Maikrouei, Ink., Kingston, Massachusetts. After the reverse step, the image of each hexagon is reproduced in the image area. The centers of the hexagons play an important role in the new image. In hexagonal centers, the true meaning of the spots depends on the number of adjacent edges. A larger number of edges gives more power within the allowed frequencies and, therefore, points with higher values. A smaller number of edges form dots of smaller values. The value of the dots is a good measure of the confidence level with synchronous reset at any given point.

(e) Determination of main axis

The hexagonal image has been restored but its orientation needs to be set.

The hexagonal honeycomb structure of the present invention has three axes spaced 60 degrees, the direction of these axes being determined by the sharpest point in the substitution area after spatial bandpass filtration. Now is the chance to see which of these three axes is the main one. This is not a necessary step. If this step is not performed, the label must be decoded three times using each of the three axes, but only one axis allows the significant message to be retrieved. The main axis is freely selected as an axis parallel to the two labeled edges as described and illustrated in FIG.

If the boundaries of the square label are determined by the data on the major axis, then most of the energy in the restored hexagonal structure will be in the middle of that square.

In determining the major axis, each of the three pretends to be the principal. The following square label drawing is determined for each axis under investigation, and the total reconstruction energy of the task structure inside this square is determined from the inverse subroutine and the digital energy output data. The right test is characterized by maximum energy. It then writes the angle of this principal axis for the initialization step and other search operations. In relation to this, it is not yet clear whether the inscribed angle is in the correct direction or 180 ° in the wrong direction. Source code lists in the microfiche attachment belonging to the main axis can be found on page 49, lines 4854, 50, lines 1-53, page 51, lines 1-5. It should be noted that there is no need to determine all three areas of the label collectively, as the energy common to the three squares does not necessarily have to be determined.

(f) Search

The recorded program SEARCH.C 300, shown in FIG. 10, combines center-centered and recorded information with recorded primary image brightness levels so that the size of each hexagonal gray level can be determined. The search shall be conducted in such a way as to minimize the possibility of loss of search. The end result is getting the size level of the gray matrix of each hexagonal data grid. Source code lists on SEARCH.C can be found on pages 52-60 of the microfiche supplement. The first part of the SEARCH.C program provides four important information grids. The CVAL grid stores a measure of the quality of the synchronous signal for each hexagon reset, while the grid GVAL allows you to memorize the gray level size (0-63) at the center of each hexagon. The remaining grids IVAL and JVAL allow you to memorize the rows and columns in the center of each hexagon.

(i) Initialization steps

On the basis of the major axis angle determined in / e / stage and the known arrangement of hexagons (5 pixels) in the example, the expected horizontal and vertical offset from the center to the other six hexagons surrounding it can be calculated by computer.

After these calculations, SEARCH.C will affect the reset signal received from the memory and the recalculated label image, also from the memory. The primary purpose of the initialization subroutine, which is found in the microfiche appendix on page 52, lines 13-14, 58, lines 1-48, page 5-47-57, page 57, lines 1-35, is the merging and condensation of information from two sources 35 Regeneration of a B-art matrix that provides the size of each hexagonal gray scale.

The search initialization step is defined by a square about 1/3 of an inch around the center of the label. Within this range, a good starting point is the point with the largest size that is in the grid reproduced by the target signal. It then determines the location of this starting point relative to the center of the label. This starting point is the point where the interrogator is powerful and different, and also a point relatively close to the center of the label. The powerful distinctive signal guarantees that the search begins with the working center of the hexagons, and preferably the point, close to the center of the label, so that its absolute position can be determined without the serious influence of distortion and clutter. The measure of a dot's quality in the synchronization reproduction structure is the size of the dot plus the size of the eight dots that surround it. The rectangular coordinates of the starting point are changed to polar form, the polar coordinates are adjusted accordingly with respect to the previously determined angle of the most important axis, and this result changes back to the rectangular shape. These coordinates are scaled in accordance with the assumed arrangement of the rows (4.5 pixels) and the columns (5 pixels) that go to the position of the hex on the hexagonal matrix. The sync quality, grayscale levels, and layouts that match the primary hexagon are then entered into the corresponding grids CVAL, GVAL, IVAL, JVAL.

ii) Basic Search Cycle

The basic search loop determines the next location of the hexagonal centers. The cycle ends when the estimated number of hexagons is found. The order of the hexagonal centers is very important. Increased reliability of the decoding process with poor labels, the LT 3516 B flies from the specific technology used below.

Each iteration of the search cycle begins with the selection of the largest synchronization point for the reproduction point, whose neighbors were not searched for their stronger values. From this known point, the search will continue for one hexagon in each of the six directions. The effect is that the search structure is raised along the trajectory from better to worse quality of the reproduced synchronization. This way, if there is a weak area of ​​the resynchronized sync, such as in the center of the label or in the blurry area, the search algorithm bypasses it instead of scrolling through it. Bypassing these weak areas and preserving them for later, the potential for network losses is reduced. Because the losses are as bad as the wrong gray reading, this search feature of the algorithm is very powerful.

The subroutine, which is in the microfiche attachment on page 53, line 5054, page 54, lines 1-53, page 55, lines 1-55, is responsible for finding the best quality master cycle sync size. The subroutine passes six times for each hexagonal neighbor of the hexagon in question. First calculates the neighbor's spacing. If this neighbor is outside the label, the cyclic iteration ends. If not, * checks the neighbor to see if it has already been found in the other direction. Cyclic iteration ends when a neighbor is searched because the algorithm makes previous searches more reliable than subsequent ones. If a neighbor passes this test, calculates the apparent location of the neighbor center in the synchronization reproduction structure. Performs a gradient search of the highest value synchronization signal at this location. Selects eight pixels surrounding the reset position to see if a higher sync value is found. If found, when the best neighbor point has eight verified neighbors for verification: is there no better value. This gradient reselection provides the degree of adaptation needed to read a folded or curled label. The subroutine then goes to the next neighbor and drills when all the neighbors have been checked.

After completing a subroutine, marks the center location so that it is not searched again. The effect is that he is singled out as a candidate whose neighbors passed the re-election. For each iteration of the cycle from 0 to 6 new candidates are added and one removed. Good tools can use a data structure that holds candidates in a queue size that allows for entry and extraction operations. One such structure is called the Priority Sequence (see Development and Analysis of Machine Algorithms; Acho, Hopkraft, and Vilman (Adison Vešli, 1974)). The linear selection algorithm is known to require n² operations when using a well-organized priority queue that uses a balanced tree or unbalanced structure requires nlogn operations. A transfer algorithm n based on group sorting can also be used if the magnitudes of the reproduced synchronization are scaled and down to a small range of integers.

g) Creating a histogram and plotting a threshold

After the end of the basic takeover cycle, the location of all the centers of the hexagons is determined, and the gray values ​​of all the hexagons that are written appear completely filled. The next step is to limit the gray level of the digital sizes in the range 0-63 to discrete levels such as black, gray and white (for black, white and gray for labels). This is accomplished thanks to the created histograms of the center of the image of the label from hexagonal centers. Quantized levels can be determined by looking for gaps in the histogram. The specific subroutine used to create the histogram and find the quantized levels can be found in the attached source code list on the microfiche appendix on page 55, lines 16-52, and on page 56, lines 1-15.

(h) Rough network adjustment and final orientation

After distorting the discrete levels, two distortions can occur. The grid may first appear off-center. This can occur if the best quality synchronization signal is positioned incorrectly at the center of the label during the initial transition. The second possibility is that the entire label is read effectively from top to bottom because the major axis angle has an ambiguity of 180 degrees.

The recorded subroutine on page 58, lines 1-54 and 59, in the microfiche attachment to lines 1-24, will allow you to perform the function of determining whether the label is centered. If the label is correctly positioned, the center row coordinates must pass through the center of the label. If a vertical layout error is made, the rows on the hypothetical center row are checked to see which one forms the line closest to the center of the label. If the row above and below is closer than the hypothetical center row, then it makes a corresponding upward or downward shift. If left-hand short alignment has been performed incorrectly, it adjusts the short-line shift by one position to the right. Horizontal layout errors and upside-down are checked using information entered in the label and known as coarse network information. The information is divided into sets of 3x3 hexagonal cells, this has already been described. Because the label can be, for example, with a grid of 33 rows on 30 columns, these sets form a grid of 11 on 10. The lower center hexagon of each complete set of 3x3 cells has a special property that is encoded. There is a guaranteed transfer from either side of the hexagon which has been discussed previously in connection with FIG. For example, if the lower center hexagon is black, the lower left and lower right hexagons must be either gray or white. The recorded subroutine on page 59, lines 27-52, and 60, in the microfiche appendix to lines 1-33, takes advantage of this transmission property to eliminate two final possible distortions. First a grid is created where each grid element indicates whether a transition has occurred between two horizontal adjacent hexagons. The grid is then inspected for each of the 9 hypothetical slides of the coarse grid arranged in structure 3x3 around the assumed slip 0. One of these slides will show the best compatibility between real and supposed transitions, and this slip position remains. The hypothesis is then tested, with the hope that the label will be read upside down. This will happen if the angle of the main axis is actually right-to-left relative to how the label was printed, not left-to-right.

If the label has been inverted, that is, the top rows have changed positions with the bottom, and the top columns with the bottom ones, then the slip results are also inverted. But there is one major change that needs to be made to correctly re-label the label. When reading short lines (length 29), the left is checked, so when the label is flipped, these labels must be checked correctly. The adjustment is exactly the procedure that will make the results of the slip hypothesis 3516 B more than a simple inversion. In fact, the best slip test results will be better than other previous tests if the label was actually read upside down.

Determining whether the label has been read upside down and whether there is any slip in the absolute position can result in the encoding of the label matrix. When the picture and slip are set correctly, the image processing functions are completed and the decoding process begins.

4. Decoding

The recorded program RD.LABEL.C 182, shown in Fig. 9 on page 61, lines 1-52 and 62 on lines 1-28 of the microfiche appendix, reads the file received by the receiving program and forms the bit sequence file containing the most suitable version. 1292 bits. Here, we use the written subroutine CELLDEC.C 183 Fig.9, on pages 63-66 of the microfiche appendix, for masking unused hexagons and utilizing a decoding program which is an inversion of the coding program.

The first step in the decoding process is to create a bit sequence from hexagonal information using the mapping hexagon - bit process, which is the inverse mapping bit - hexagon process, using the encoding operation.

The bit (information) sequence subsequently duplicates the program into a high priority message bit sequence and a low priority message bit sequence or as many bit sequences as are used to encode the label.

After that, it is necessary to apply error correction for each bit sequence using the error encoding technology used in the label encoding proLT 3516 B process. For example, if Reed-Solomon coding is used, with error correction, the bit sequence that is created by the receiving program generates an output signal that exists in the format described above for file encoding. The correction of errors occurs in the following sequence (see: Theory and practice of error control codes described above):

1. Calculate syndromes

2. Calculate the polynomial of error search using Berlekamp-Mesi algorithm

3. Calculate the location of the error using the Chen transition

4. Calculate the magnitude of error using the Forney algorithm.

The last step is used if the number of errors to be corrected in steps 2 and 3 is detected. Also calculates the number of detected errors. If an undetectable number of errors is detected or if the error is in the meaningful part described earlier), a flag is placed. The specific coding error procedure used in the above example is in the microfiche appendix on pages 67-75 and is noted in Figure 9 of ERRD EC.C 184.

5. Output

When scrolling the cargo (identifying its location on the conveyor), high priority messages pointing to the postal code of the location can be used to activate the appropriate guide levers, or the conveyor to direct the cargo to the appropriate truck, airplane or freight wagon carrying the cargo in the required direction.

used in range,

Although. the invention may be utilized in a conveyor belting system, obviously it may be in large operations information collection and processing, preferably reading a label on a cargo, letter, part, machine, or similar device and forcing the system to process or perform manufacturing operations, e.g. on a product bearing a label. The present invention enables these operations to be performed quickly and with great precision. Speed ​​and accuracy are associated with a substantial amount of label information, and even allow you to protect a large amount of information from loss due to label breakage or other damage.

As shown in FIG. 9, the TEXTOUT.C 185 program may be used in the computer terminal for the decoded message. The TEXTOUT.C program can be found on pages 76-78 of the microfiche supplement.

Inscriptions on drawings

FIG. 6th

208 ... transducer shaft, 202

204 ... microprocessor, 206 coil height sensor

FIG. 7th

1. Label lighting

2. Optical measurement of reflected image

3. Reflected image processing

a) Finding the target (optional)

(b) Normalization of measured image (optional)

c) Scaling the image (optional)

d) Two-Dimensional Synchronous Recovery I Edge Enhancement Operation

II Framing (optional)

III Two-Dimensional Fast Fourier Change

IV Image Filtering

V Reverse Fast Fourier Change

(e) Determination of main axis (optional)

(f) takeover

I Initialization a

II Basic Takeover Cycle

(g) histogram generator (optional) coarse grid determination and final orientation (optional)

4. Decoding

5. Output

FIG. 8th

190 ... crossed analog threshold

191 ... perform coarse 2-D correlation

192 ... successful?

193 ... to perform the correct inspection

194 ... successful?

195 ... center of label found