Method for correcting the image data of a camera system

The invention relates to a method for correcting the image data of a camera system according to the preamble of claim 1.

find particular in the printing industry camera systems for different Applications are becoming common. For example, such Camera systems as a register measuring systems, inspection systems or Find web inspection systems use. are thereby often as color camera digital CCD cameras used, the photosensitive pixels in accordance with the sampled color in the observation area three outputs for three separate Color channels, mostly for the colors red, green and blue, provide.

A problem of the known camera systems for the inspection of colored Printed material is that supplied by the color camera image data often not correspond to the color perception of the human eye. Unedited image data these color cameras are in terms of color balance, brightness, contrast and Tone reproduction insufficient in the color tuning, the human color perception corresponds. In addition to the shortcomings of lenses and lighting devices, the spectral sensitivity distribution of Color cameras used main reason for this problem. The sensitivity distribution the color cameras used does not match the sensitivity distribution of the match human eye, which leads to that supplied by the color cameras Image data in the downstream further processing, for example, the display of a monitor, lead to a distorted visual impression.

From US 5,189,511 A is a method for correcting image data by means of a Correction matrix known.

The US 4,930,009 A, US 5,331,441 A and US 6,278,533 B1 disclose methods for correcting image data of a camera system with a color camera under Using a correction matrix.

The invention is based on the object, a method for correcting the image data of Camera system to provide.

The object is achieved by the features of claim 1.

Prerequisite for a sensation proper processing of the image data of the Camera system is that as many colors captured by the color camera terms hue, saturation and brightness sufficiently accurate in a color space are classified corresponding to the color perception of the human eye. Particularly suitable is the so-called CIELAB color space, the pressure in the art became widespread. A measure of the accuracy of sensation according to color differences is in the CIELAB color space by the geometric given distance of the nominal and actual values ​​of L, A and B (.DELTA.E).

An achievable with the invention advantage is in particular that by the Multiplying the output vectors with a particular square Correction matrix, the image data can be changed in a simple manner such that they are approximated to the color perception of the human eye. Especially achieved by this multiplication by a correction matrix for a relatively accurate Classification of all inks in a fundamentally arbitrary color space. It should also the conversion by multiplying it with the correction matrix data technology so easy to realize that even with large amounts of image data, an implementation in real System is possible.

Of crucial importance for the quality of the correction of the present invention Image data are, of course, the coefficients of the correction matrix, because depending on the choice these coefficients transforms the output vectors in different ways will. The coefficients of the correction matrix can be made of, for example, be determined experience and then permanently stored. To the coefficients of Correction matrix variably to different boundary conditions, for example, in color camera, the lighting conditions or the used to adapt the optics used, is an iterative approximation algorithm proposed. To carry out this approximation algorithm has a Reference color chart, for example, an IT8 chart with 288 color patches can be specified. In the color fields of the different reference colors are shown. It should also the classification of the various reference colors in a suitable color space, For example, the CIELAB color space known. let by known transformations from these predetermined CIELAB values ​​for the various reference colors calculate reference color table corresponding desired outputs for the three color channels. As a result, that is, a reference color table is used for the approximation algorithm as Input and for each reference color, a target vector for the three color channels as desired result of the conversion set. In carrying out the Approximation algorithm for determining the coefficients of the correction matrix, will now be Reference color chart included with the color camera and for each color field of Output vector of the color camera determined. The difference between these corresponding output vectors of the color camera and the predetermined nominal vectors the difference between the color perception of the human eye and the Sensitivity distribution of the color camera.

Another advantage of this method is the calculation of the color correction values ​​for different illumination sources and changes of the same. In the printing technology is currently still the standard light D50 uses. is by setting the white point D50 it is possible that 709 adapt Rec. by a conversion to the D50 standard light so that the intensities of the non-linear R ', G', B 'values ​​behave as if to object under examination with a D50 lighting was illuminated. Through a metrological method that iteratively, the values ​​of R ', G', B 'color space to the CIELAB color space adjusts, it is possible to adjust the color spaces to each other without the real standard lighting is necessary. This method has the advantage that when an expected change in the standard light setting immediately adapt can be made.

The starting point of the iteration is a correction matrix whose coefficients as Output values ​​are given. These outputs may be either purely coincidental or according to certain empirical values ​​selected. In the first iteration will be this correction matrix multiplied by all output signal vectors and the characterized cached obtained corrected output signal vectors. Then the coefficients of the correction matrix are changed slightly and the Multiplication performed again. The change in the coefficients of the correction matrix is in each case only be accepted if the corrected output signal vectors approach the predetermined nominal vectors.

The approach of the corrected output signal vectors to the predetermined Target vectors must be evaluated for each iteration to use this to vote to decide whether the action taken in this iteration The coefficients of the correction matrix accept or reject them shall be. A particularly suitable assessment procedures represents it there when for each color patch of the reference color chart, the difference value between the corrected Output value and determined the preset for this swatch target vector and the sum of all these difference values ​​is added. The change of Correction coefficients of the correction matrix in the last iteration is then only for the assumed case where the sum of all the difference values ​​in the last iteration step in the become compared to the sum of all difference values ​​in the penultimate iteration small is. However, if the sum of all the difference values ​​by changing the coefficients of the become larger correction matrix in the last iteration, the change is Coefficients discarded. on This summary consideration of the difference values all reference colors, it is quite possible that the difference for individual Reference colors during an iteration increases. Overall, however, is reliably minimizing the difference values ​​of all color channels backed away.

Another problem of existing camera systems is the right attitude of Color balance, that is, the correct weighting of the three color channels to one another. To the Color balance of each color channel to adjust relative to each other, can each output vector added a correction vector and at the same time Coefficients of each output vector with three color channel-dependent Correction factors are multiplied. This correction of the three color channels each Output vector corresponds to a linear mapping of the different coefficients the output vectors.

A particularly good color balance is achieved when the correction vector and the three color channel-dependent correction factors are chosen such that they by Application of the correction with the correction vector and the three correction factors obtained corrected output signal vectors for the two fields with the Reference gray values ​​black and white substantially exact for these two Swatches meet predetermined nominal vectors. D. h. In other words, the linear mapping of output vectors is selected so that both the Reference gray values ​​black and white corrected results revealed that the Contrast the human eye would correspond. This linear transformation is applied to all output signal vectors, whereby brightness and contrast complete color spectrum are automatically also corrected.

To carry out the process of the invention are particularly CCD color cameras suitable with a plurality of flat or linearly arranged pixels. These CCD color cameras also have three color channels and deliver the image data pixel by pixel Output vectors whose coefficients respectively, the three output signals for the three Color channels (red, green, blue) represent. Is the color CCD camera, for example, with 1 provided million pixels, this corresponds for each image of an image dataset of 1 million Output vectors with three coefficients.

When using CCD color cameras can cause color distortion and a Drop in intensity especially coming on the edges of the camera images. These Distortions generated by the objects used. To correct this Intensity drop may be a so-called shading correction are used. To be specified for each pixel three color channel-dependent correction factors. By Multiplying these pixel dependent correction factors with the coefficients of Output vectors, the pixel-specific color distortions or a design-related drop in intensity in different areas of the CCD chip be compensated.

This pixel-specific, color channel-dependent correction factors, for example, in easily be determined experimentally by the fact that the observation area the CCD color camera with a homogeneous material, in particular homogeneous white Material, designed and by triggering the camera for each pixel Output vector is determined. is for all these output vectors then the output vector filtered out, which has the highest coefficient and thus the brightest spot represented in the observation area. Since the is the observation area but designed with a homogeneously colored material would all pixels substantially identical to matching one another Output vectors deliver. Thus, the respective differences are based on Color distortion, or a maximum design intensity drop. to this compensate, correction factors are now for each color channel of each pixel selected, which ensure that all in view of the homogeneous colored material Output vectors to the output vector at the brightest spot in the Observation area meet.

In particular, color distortions are strongly dependent on the lighting conditions in Observation area from. To errors by changing the lighting conditions exclude, therefore, should the lighting in the experimental determination of pixel specific color channel-dependent correction factors of lighting during the Later use of the camera system correspond.

In many applications of the inventive method, the corrected Output vectors of the correction by the output vectors Color camera can be obtained for driving the three separate color channels of a Colour monitor used. raises Color appearance of a Colour Monitor likewise the problem that the display characteristic of most Colour monitors does not match the color the human eye. This due in particular to the fact that the brightness behavior of monitors usually is not linear, ie, the intensity of the light that will be reproduced on the screen is a non-linear function of the electric input signals. This means in other Words, for the case that the present invention corresponding to the color perception the human eye corrected output signal vectors easy to transmitted Colour Monitor and there without considering the non-linearity of are brightness behavior displayed unwanted distortions in Colour on the screen may occur.

To such color distortion when viewing on a Colour Monitor to prevent the coefficients of the corrected output signal vector can be used as base each with a factor γ are set to the power. Due to this non-linear Conversion of the coefficients of the corrected output signal vectors, the offset nonlinearity of the brightness behavior of most Colour Monitors will. For most color image monitors the factor γ has thereby in the range between 0.3 to 0.5, in particular about 0.45, can be selected.

To the source of illumination when using corresponding camera systems not on a to calibrate standard light source, can according to the invention, a further Correction step are performed. In this correction step, the coefficients are the output vectors so converted that the outcome of the Output vectors corresponding to the part in illumination of the observation area a standard light would get.

An embodiment of the invention is illustrated in the drawings and will in described hereinafter.

The single drawing shows the various method steps in carrying out one embodiment of the inventive method.

With an opening formed in the manner of a CCD camera color camera 01 is an in Observation area 02 is arranged, color printed printed product 03 added. In the color camera 01, a CCD chip is provided, the converts image information in the observation area 02 into electronic image data. at this reaction is one of each of the photosensitive pixels of the CCD chip Output vector 04 generated. According to the number of pixels on the CCD chip be by the color camera 01 according to many output signal vectors 04 to Further processing provided.

Each output signal vector 04 is formed by three coefficients R, G and B. The Coefficients R, G and B corresponding to the color values ​​for the three color channels red, green and blue, the color of the printed product 03 at the position in Observation area, which was taken from the corresponding pixel, the mixture red from the three color channels, green and blue corresponding.

The output signal vectors 04 whose index of arrangement of the respective pixel on the CCD chip corresponds to the raw data in a first correction module 06 for Adjusting the color balance of the brightness and contrast further processed. To the coefficient R, G, B of the output vector are 04 all with the color channel-dependent correction factors K₁, K₂ and K₃ multiplied and the resulting resulting resultant vector a correction vector 07 with the fixed value coefficients a₁, a₂and a₃ added. By this arithmetic operation, the corrected Output vectors 04a generates the color balance, brightness and contrast improve the image data. This object is achieved in that the color channel-dependent correction factors K₁, K₂ and K₃ and the coefficients a₁, a₂ and a₃the correction vector are 07 selected such that when recording the reference gray values Black and White produced the case of the color camera (01) Output signal vectors 04 are transformed such that the corrected received Output vectors 04a such setpoints correspond, as from the Conversion of the known CIELAB color values ​​in target vectors results.

Then, the corrected output signal vectors 04a are at a second Correction module 08 passed. In correction module 08, each output vector 04a multiplied by a 3x3 correction matrix 09 and from the corrected Output vectors 04b calculated. The coefficients K₄ to K₁₂ the correction matrix 09 were doing previously determined in a suitable iteration such that the in the output signal vectors 04a contained image information to the color perception the human eye can be approximated.

Then, the corrected output signal vectors 04b to a third Correction module 10 passed. In the third correction module 10 are in a database each pixel color channel-dependent correction factors stored which the adapting dependent on the position of the respective pixel intensity values ​​by the coefficient R, G, and B are multiplied. In the result, the corrected output signal vector 04b of the first pixel by the correction factors K₁₃, K₁₄ and K₁₅ multiplied to derive for the first pixel a corrected output signal vector 04c to calculate. The Correction of the corrected output signal vectors 04b is carried out pixel by pixel. The therefore the number of pixel-specific correction factors corresponding to three times the Number of pixels.

The corrected output vectors are then 04c to a fourth correction module 11 passed. In the fourth correction module 11, the coefficients of R, G, B of corrected output signal vectors 04c potentiated by a factor γ and from the corrected output signal vectors 04d calculated. By potentiating the Factor is γ the nonlinear brightness transfer function of a monitor 12 considered, the 04d, the corrected output signal vectors for display be transmitted.

As a result, by correcting the output signal vectors 04 in the Correction modules 06, 08, 10 and 11 reaches that of the screen Colour monitor 12 displayed color images to color the human Eye are adapted so that the visual impression when viewing the display on Colour Monitor 12 well corresponds to the color perception, which in the immediate viewing the printed product would arise 03.

LIST OF REFERENCE NUMBERS

<dl tsize="13" compact="compact"><dt>01</dt><dd>color camera</dd><dt>02</dt><dd>observation area</dd><dt>03</dt><dd>Printed Media</dd><dt>04</dt><dd>Output vector</dd><dt>04a</dt><dd>corrected output vector (first correction step)</dd><dt>04b</dt><dd>corrected output signal vector (second correction step)</dd><dt>04c</dt><dd>corrected output signal vector (third correction step)</dd><dt>04d</dt><dd>corrected output signal vector (fourth correction step)</dd><dt>05</dt><dd>-</dd><dt>06</dt><dd>first correction module</dd><dt>07</dt><dd>correction vector</dd><dt>08</dt><dd>second correction module</dd><dt>09</dt><dd>correction matrix</dd><dt>10</dt><dd>third correction module</dd><dt>11</dt><dd>fourth correction module</dd><dt>12</dt><dd>Colour Monitor</dd><dt>R, G, B</dt><dd>Color channel coefficients (output vector)</dd><dt>K₁, K₂, K₃</dt><dd>color channel-dependent correction factors</dd><dt>K₄ to K₁₂</dt><dd>Coefficients of the correction matrix</dd><dt>K₁₃, K₁₄, K₁₅</dt><dd>color channel-dependent correction factors of the first pixel</dd><dt>K₁₆, K₁₇, K₁₈</dt><dd>color channel-dependent correction factors of the second pixel</dd><dt>a₁, a₂, a₃</dt><dd>Fixed value coefficient of the correction vector 07</dd><dt>γ</dt><dd>Potenzierungsfaktor</dd></dl>