Evaluation of the relief of a tyre surface by active stereovision

The present invention relates to the field of visual inspection of tires.

Conventionally, such an inspection requires the skill of an operator, and the operator visually inspects the tire in order to detect a condition considered to be a visible defect on the surface of the tire. These tasks are time consuming and costly, which is why manufacturers are actively looking for means to assist operators.

In this connection, it has been found that it is necessary to collect a digital display of the unevenness of the tire surface to be inspected, the purpose of which is to use this digital display as a reference for the tire surface after analysis and processing. It is to compare with the image or the data obtained from the model. This digital display of the surface is also called the surface unevenness image.

In particular, the present invention relates to the field of stereoscopic collection of uneven images on the surface.

Various image collection methods are disclosed for the purpose of providing as appropriate data as possible to a digital processing means capable of comparing this image with a reference image in order to determine the conformity of the tire to be analyzed.

It has been proposed to use old-fashioned stereoscopic techniques to use two separate cameras, each dedicated to collecting data on unevenness and collecting data on appearance, such as color, grayscale or brightness.

This solution, called the passive stereoscopic method, requires the images coming from the two collection means to be associated. This correspondence may be determined using the presence of characteristic elements of the image, such as characteristic corners or contour shapes. Next, the coordinates of the surface are calculated by the triangulation method by finding the angles of different figures of the same point on the surface seen by the two cameras.

However, some assumptions are needed to properly execute the computational algorithm. This is because ambiguity may occur if the surface to be evaluated has a light reflection or light refraction region. In this case, the algorithm cannot accurately determine the correspondence between the pixels of the two cameras. In addition, unlike the human brain, the algorithm does not have knowledge of the topography or context of the image to be analyzed. Therefore, it may be necessary to involve the operator in the analytical process to select the points that should be associated.

For this reason, in contrast to passive optical technology, collection techniques called active techniques have been developed that send optical signals at various angles onto the surface to be reconstructed that is visible to the camera. .. Its purpose is to facilitate the correspondence of surface points.

In these methods, the surface is illuminated with known light emitting features detected by the camera's light receiver. The task of associating the images recorded by the two stereoscopic cameras is facilitated by knowing the feature elements, and thus the ambiguity described above is resolved during the analysis.

As described in detail below, one of the most commonly used structured light projection algorithms is formed by a series of dual features consisting of alternating illuminated and non-illuminated lines. Illuminate the surface with light. At the same time, the camera collects these series of consecutive images, each of which is illuminated or unilluminated at a point on the surface. It is then possible to reconstruct the alternating illuminated and non-illuminated bands visible by the two cameras and identify the optical bands in a one-to-one manner, the purpose of which is the location of a point on the surface. The purpose is to reconstruct the uneven image on the surface by making the images from the two cameras correspond to each other.

Proper use of these lighting algorithms collects images of the tire surface while avoiding the effects caused by shaded areas when the surface irregularities are significantly chopped, but the image processing means also stains or changes color. It is possible to provide sufficient information to identify the resulting luminance effect.

In applying the above method to the evaluation of tire surface irregularities, if it is desirable to optimize the image acquisition cycle, especially if it is desirable to reveal the tread irregularities, if some modifications are required. There is.

FIG. 1 shows the case of a conventional application. In this conventional application, the lighting means 20 projects a fringe system onto the tread, the stereoscopic cameras 10a and 10b are emitted by the lighting means 20 (E), and the tire P. Arranged to collect the (F) light reflected by the surface of the tire. The tires are attached to the rim 30 of the wheel 31 which is rotated about axis D by a motor driven support hub 32.

At each acquisition pass, the camera records two stereoscopic images of the angular portion α of the surface of the tread. A complete image of the tread is obtained by rotating the tire around its axis of rotation D and matching the 2π / α photographs taken by each of the stereoscopic cameras against each other.

Further, in the concrete example of the algorithm, it is necessary to continuously project the fringe system S of the format shown in FIGS. 2, 3 and 3a one after another. The fringe system consists of alternating illuminated and non-illuminated bands (S1, S2, S3, S4) of known width according to a predetermined binary code, and these fringe systems are recorded by the camera. It is related to coding and decoding techniques that can identify the fringes of the projected image.

The stereoscopic camera collects the projected images of each of the continuous fringe systems S1, S2, S3 and S4 on the surface of the tire.

With reference to FIGS. 3a and 3b, the fringe system S1 corresponds to the first column. The fringe system S2 corresponds to the column 2, the fringe system S3 corresponds to the column 3, and the fringe system S4 corresponds to the column 4. The number of fringe systems that can be projected is, of course, not limited.

Processing systems, including known algorithms (the description of these algorithms does not form part of this specification), associate continuous lighting levels with each point on the tire surface to resolve positioning uncertainty. Decode the image for.

Therefore, in the first method, the fringe system is continuously projected onto a portion of the tire, and then the tire is rotated about its axis to iterate over the continuous angular sectors. In the second method, each fringe system is imaged over the entire rotation of the tire, and the same number of rotations as the fringe system to be projected is performed.

Whatever method chosen, these continuous rotations have been found to be time consuming and reduce the efficiency of the inspection system. Also, such solutions require particularly accurate coding and synchronization means.

Fringe systems to be projected in connection with stereoscopic means for collecting digital uneven images of the tire surface, as proposed in U.S. Pat. No. 4,175,862 to reduce collection time. It is possible to arrange the same number of fringe projection devices as the number N of.

In this way, by rotating the tire once around its axis of rotation, it is possible to collect 2N images of the entire surface of the tire illuminated by the N fringe system, which came from the 2N stereoscopic camera. ..

<p><patcit num="1"><text>U.S. Pat. No. 4,158,662</text></patcit></p>

<p> However, such devices require a large number of cameras and projectors, which can interfere with each other and many to align the N uneven images on the surface with each other. It may have the additional drawback of requiring additional calculations.</p>

<p> The collecting device for taking a digital uneven image of the tire surface of the present invention is -Has two color cameras for stereoscopic image collection, each camera has N primary image sensors for a given primary color, N is 2 or more, and the primary image sensor is Arranged to collect the light emitted by the lighting means towards a predetermined area of the tire surface and reflected by the tire surface. -The tire surface according to a fringe system consisting of alternating illuminated and non-illuminated bands of a given width, each independently and along the same direction light with a wavelength corresponding to one of the camera's primary colors. It has N lighting means that project simultaneously on the area of.</p><p> For the same angular sector, each camera takes a set of fringes at the same time, so as will be understood later, a given point on the tire surface is illuminated with one fringes. It is not very important that another fringe system is considered unlit.</p><p> Therefore, one rotation of the tire around the axis of the tire gives an image of N pairs of the circumferential surface of the portion of the tire that should be analyzed, given that each pair is placed within each of the color cameras. Obtained from two sensors of color. These N pairs of images form N stereoscopic images of N fringe systems.</p><p> The present invention also has the advantage of reducing the number of image alignment and camera calibration tasks. Because the pictures are taken at the same time. Similarly, this collection mode avoids the problem of light interference caused by the illuminating means.</p><p> The following description helps to better understand the collection system of the present invention and is supported by FIGS. 1-6.</p>

<figref num="1">It is a schematic diagram of the conventional stereoscopic visual inspection apparatus.</figref><figref num="2">It is a photograph of the surface of the tread illuminated by the fringe system.</figref><figref num="3a">It is a figure which shows the fringe system structured according to the binary code.</figref><figref num="3b">It is a figure which shows the fringe system which was structured according to the Gray code.</figref><figref num="4">It is a schematic diagram of the apparatus of this invention.</figref><figref num="5">It is a figure which shows an example of the wavelength distribution spectrum of the primary color used in a color monitor.</figref><figref num="6">It is a schematic diagram explaining the principle of the color camera used in the stereoscopic image collecting means of this invention.</figref>

As mentioned above, the device shown in FIG. 1 schematically illustrates the stereoscopic image collecting means 10 formed by the two cameras 10a, 10b, where each camera has a tire P to be inspected. It comprises an incident objective lens into which a predetermined region Z of the surface, in this case the reflected light F coming from the tread, is incident. This surface is illuminated by a lighting means 20 capable of projecting one or more fringe systems, consisting of alternating illuminated and non-illuminated bands, onto the surface of the tire as seen by the objective lens of the camera.

FIG. 2 is a photographic view of the above-mentioned region Z of the tread of the tire P illuminated by the fringe system. The bright lines are parallel to each other and are preferably located in the circumferential direction.

In relation to the embodiment of the present invention, it is also assumed that the lines constituting the fringe system are arranged in the lateral direction and the radial direction, or that a concentric line system is formed particularly when the surface of the sidewall of the tire is analyzed. it can.

FIGS. 3a and 3b show the case of the fringe system in which the width of the fringe system bands is inversely proportional to the number of bands. Thus, the width of the fringe S2 band is equal to 1/2 the width of the fringe S1 band, the width of the fringe S3 band is equal to 1/2 the width of the fringe S2 band, and so on. The same is true. In other words, the width (L1, L2, ... LN) of each band of the fringe system (S1, S2, ... SN) is the smallest bandwidth L.₄Multiple of the width of the fringe band with (modulo 2)ⁿ), N is 1 to (N-1), and N is equal to -4 in the examples shown in FIGS. 3a and 3b.

Thus, the larger the number N of fringe system ranks, the narrower the band width and the higher the analysis accuracy, but the longer the collection time and calculation time. In addition, the width of the fringe is limited by the resolution of the camera sensor.

Determining whether a pixel observed by a camera is illuminated has been found to be difficult due to the mutual reflection between the surface to be inspected and its environment. In addition, the binary coding scheme causes an accumulation of errors at the boundaries of the bright bands. Specifically, in the binary code as shown in FIG. 3a, the boundary between lines 8 and 9 exists in all fringe systems.

Figure 3b shows Hall-Holt and Rusinkiewicz, proposed by Gray (Bell Laboratories, 1953) and presented at the 2001 International Conference on Computer Imaging. ), Or non-patent literature, Lucin Kivics, Hall-Holt and Levoy, "Real-time 3D model acquisition", Proceedings of. It shows the specific code given in SIGGRAPH 2002 (Proceedings of SIGGRAPH 2002), Volume 1, July 2002, pp. 438-446. This particular code consists of illuminating the surface with a system of bright fringes, where the width of the fringe is also halved at each continuous image, but at each boundary between the two bands. , I can only see it once. This device describes how analytical errors that can occur in the boundary region can be reduced.

Other codes and other reconstruction algorithms, such as those proposed by Hall-Holt and Lucinkivics, have been proposed to increase collection and analysis reliability, but these various yielding structured images. Explain the usage of the means, and more appropriately, these various means should be evaluated using a fringe system consisting of illuminated and non-illuminated bands that are essentially alternated according to a given code. It is not intended herein to explain that it consists of illuminating a surface.

To make the best use of the results, the fringe systems projected by each of the lighting means are arranged so that the boundary regions are aligned in such a way that they are aligned.

The apparatus of the present invention is shown in FIGS. 4, 5 and 6.

For this collector, it is believed that the number of fringe systems to be projected to obtain sufficient accuracy is small. In fact, projection of three or four fringe systems is sufficient, and it is conceivable that the object of the present invention takes advantage of this situation.

The apparatus of the present invention has a stereoscopic image collecting means formed by two color cameras 13a and 13b.

In the technology currently in use, this type of camera is a means by which reflected light from an object for which it is desirable to collect an image can be divided into a certain number of basic colors (R, G, B). Has.

These dividing means are preferably formed by a filter consisting of primary color cells, well known as a pair of prisms or Bayer filters. The function of these means is to divide the light into a certain number of colors called base or base colors. In general, such filters divide the light into three base or basic colors: red (R), green (G) and blue (B). However, it is also possible to create a camera that contains four or more basic colors. As an example, cameras are on the market that include four basic colors: red (R), green (G), blue (B) and cyan.

Therefore, the reflected light F coming from the object to be inspected is decomposed into as many monochromatic zones as there are base or base colors. Each of these images was then directed to a light-sensitive photosite, eg, a particular sensor formed of an assembly of CCD or CMOS sensors, which these photoelectric photosites received. The amount of light can be converted into current. Therefore, the same number of grayscale images as the existing basic colors can be obtained. The maximum resolution of the sensor is determined by the number of pixels forming the final image and the number of corresponding photosites.

The present invention comprises utilizing this mode of operation of a color camera to obtain specific information about surface irregularities to be evaluated.

For this purpose, referring to FIG. 4, the N stereoscopic image collecting means that have come to collect 2N images of the surface of the tire illuminated by the N fringe system are two color cameras 13a, Formed by 13b of 2N primary image sensors, two sensors of the same given primary color of each of the cameras form a stereoscopic image collection means.

In this case, it is sufficient for each of the illuminating means to illuminate with light having a wavelength corresponding to one of the primary colors of the camera according to a given fringe system, the purpose of which is to illuminate two N fringe systems with two. The purpose is to enable simultaneous and separate viewing by the primary color sensor of the camera.

In this case, the maximum number N of fringe systems that can be projected onto the surface corresponds to the number N of the primary colors of the camera.

FIG. 6 shows the operating principle of one of the color cameras (13a) forming the stereoscopic image collecting means. The operational details of the related color camera 13b, which can use the subscript b instead of the subscript a, are not shown because they are the same.

A ray of incident light F enters the camera and illuminates the reflecting prisms 134a (134b), 135a (135b) and 136a (136b), respectively, which divide the light into base or basic colors and divide this light. The light is reflected and directed to the luminance sensors arranged in the camera, that is, the luminance sensors 131a (131b), 132a (132b), 133a (133b), respectively, and these luminance sensors form an image of the surface. Can be done.

These colors are the base or basic colors as shown in FIG. 5, in which case blue (B) corresponds substantially to a wavelength of 450 nm and green (G) corresponds to a wavelength of 550 nm. However, red (R) corresponds to a wavelength of 680 nm.

Therefore, the illumination means (231) has a wavelength of 450 nm, which corresponds to blue, and this fringe system is illuminated with alternating blue light to make it visible by sensors 131a (131b) assigned to this color. It suffices to configure the first fringe system S1 consisting of a band and a non-illuminated band. The second fringe system S2, which is different from the first fringe system, is simultaneously emitted by the illumination system (232) at a wavelength of 550 nm and is therefore visible only by the green-only sensor 132a (132b). This fringe system consists of alternating bands illuminated by green light and non-illuminated bands. The third fringe system S3 emitted by the illumination system (233) at a wavelength of 680 nm is visible by the sensor 133a (133b) dedicated to red light, and such a third fringe system is visible in alternating red bands and non-red bands. Has a lighting band.

Thus, even with only two color cameras, by separately collecting the images coming from each of the three sensors of the color camera, they are simultaneously emitted by the illuminator according to different wavelengths that correspond to the base or base color of the camera. It is possible to obtain stereoscopic images of three different fringe systems S1, S2, and S3.

The blue primary color sensor 131a of camera 13a is associated with the blue primary color sensor 131b (not shown) of camera 13b. These two sensors form a collecting means for taking a stereoscopic image of the tire surface illuminated by the fringe system S1 emitted by the blue-corresponding lighting means 231. The green primary color sensor 132a of camera 13a is associated with the green primary color sensor 132b (not shown) of camera 13b. These two sensors form a collecting means for taking a stereoscopic image of the tire surface illuminated by the fringe system S2 emitted by the green-corresponding lighting means 232. The red primary color sensor 133a of the camera 13a is associated with the red primary color sensor 133b (not shown) of the camera 13b. These two sensors form a collecting means for taking a stereoscopic image of the tire surface illuminated by the fringe system S3 emitted by the red-corresponding lighting means 233.

Therefore, the two color cameras 13a and 13b can see the three fringe systems at the same time, and collect the image of the entire surface of the tread illuminated by the three fringe systems once on the tire around the rotation axis D of the tire. It can be done by rotating.

A given point on the surface of the tire can be illuminated simultaneously with two different colors, such as blue and green, which are illuminated in the fringes S1 and S2 and not in the fringes S3. Be considered.

To avoid distortion caused by light being projected onto the tire surface at different angles, the illuminating means 23 each illuminate the tire surface at a given wavelength according to a given fringe system. It is composed of three lighting means 231,232,233 that can be used. Thus, means 231 emits a first fringe system S1 at a wavelength corresponding to blue (B), means 232 emits a second fringe system S2 at a wavelength corresponding to green (G), and means 233. Emits a third fringe system S3 at a wavelength corresponding to red (R). These three fringe systems are emitted simultaneously and are directed to the tire surface at the same given angle using a semireflector 234.

In order to limit the parasitic effect caused by the wavelength corresponding to near infrared rays, it was found that it is advantageous to place the filter on the objective lens of the camera which can limit the incident of light rays having a wavelength exceeding 750 nm. ing. Other band filters may be added to block the incident of light of undesired wavelengths. Thus, in general, the purpose of this filter is to block the passage of light rays having a wavelength different from the wavelength used for the selected illumination.

In order to embody the stereoscopic image collector of the present invention described above, it has been found to be advantageous to collect additional (N + 1) images of the tire surface to be evaluated. The purpose is to automatically determine the calibration threshold to distinguish the illuminated band from the non-illuminated band. To this end, N images are created by continuously illuminating the entire surface of the tire and illuminating the fringes using each of the lighting means corresponding to each of the base or basic colors, and illuminating. One additional image is formed in which is not done at all.