Target object surface quality detection method, device, system, equipment and medium

Abstract

Embodiments of the present specification provide a target object surface quality detection method and device, system, equipment and medium, wherein the detection method comprises: acquiring a plurality of images of a target object, comprising: controlling respective regions of the target object to move to a preset reference position respectively, and acquiring images of the target object located at the reference position to obtain images of the target object covering all regions; determining overlapping regions in each adjacent image according to coordinates corresponding to the reference position in each adjacent image in the images of all regions covered by the target object and field of view size information of each image; calculating a relative offset parameter of the overlapping region in each adjacent image respectively; splicing each adjacent image according to the relative offset parameter to obtain a target image; and detecting the target image to obtain surface quality data of the target object. The above scheme can improve detection efficiency.

Description

TECHNICAL FIELD

[0001] Embodiments of the present specification relate to the technical field of image processing, and particularly relate to a target object surface quality detection method and device, system, equipment and medium. BACKGROUND

[0002] In the field of production and manufacturing, for some high-precision products (for example, wafers), after completing the corresponding processing process, the surface of the product needs to be detected to determine whether it meets the quality requirements.

[0003] In actual operation, in order to obtain the complete surface of the entire detected product, the same measured product needs to be photographed multiple times, and then the multiple images obtained by photographing are spliced to obtain a complete panoramic image, and the spliced panoramic image is detected.

[0004] At present, in the entire detection process, when splicing the multiple images obtained by photographing, since the photos are randomly taken, it is difficult to determine the overlapping area between adjacent images, and all points (including overlapping areas and non-overlapping areas) on the multiple images need to be processed, and then the processed multiple images are spliced.

[0005] Therefore, using the existing scheme, a large amount of time is needed to determine the overlapping area between adjacent images, and the image splicing takes a long time, resulting in low detection efficiency. SUMMARY

[0006] Therefore, the embodiments of the present specification provide a target object surface quality detection method and device, system, equipment and medium, which can improve the detection efficiency.

[0007] Firstly, the embodiments of the present specification provide a target object surface quality detection method, comprising:

[0008] obtaining multiple images of a target object, comprising: controlling the corresponding regions of the target object to move to a preset reference position respectively, and obtaining the images of the target object located at the reference position to obtain the images of the target object covering all regions;

[0009] determining the overlapping area in each adjacent image according to the coordinates corresponding to the reference position in each adjacent image in the images of the target object covering all regions and the field of view size information of each image;

[0010] respectively calculating the relative offset parameters of the overlapping area in each adjacent image;

[0011] splicing each adjacent image according to the relative offset parameters to obtain a target image;

[0012] detect the target image to obtain the surface quality data of the target object.

[0013] Correspondingly, the embodiments of the present specification also provide a target object surface quality detection device, comprising:

[0014] an image acquisition unit adapted to acquire a plurality of images of the target object, wherein the image acquisition unit is adapted to acquire images of the target object at a preset reference position by controlling respective regions of the target object to move to the preset reference position respectively, to obtain images of all regions covered by the target object;

[0015] an image determination unit adapted to determine overlapping regions in each adjacent image according to coordinates corresponding to the preset position in each adjacent image among the images of all regions covered by the target object and field of view size information of each image;

[0016] a parameter calculation unit adapted to calculate a relative offset parameter of the overlapping regions in each adjacent image respectively;

[0017] an image splicing unit adapted to splice each adjacent image according to the relative offset parameter to obtain a target image;

[0018] an image detection unit adapted to detect the target image to obtain the surface quality data of the target object.

[0019] The embodiments of the present specification also provide a target object surface quality detection system, comprising:

[0020] an image acquisition device adapted to acquire images of the target object;

[0021] The target object surface quality detection device of any one of the foregoing embodiments is adapted to detect the images of the target object.

[0022] The embodiments of the present specification also provide a target object surface quality detection device comprising a memory and a processor, wherein the memory is adapted to store one or more computer instructions, and the processor executes the computer instructions to perform the target object surface quality detection method of any one of the foregoing embodiments.

[0023] The embodiments of the present specification also provide a computer readable storage medium having computer instructions stored thereon, wherein the computer instructions perform the target object surface quality detection method of any one of the foregoing embodiments when executed.

[0024] The target object surface quality detection method provided in this specification's embodiments, by controlling the corresponding areas of the target object to move to preset reference positions, can acquire images of the target object located at the reference positions, thereby obtaining images covering all areas of the target object. Based on the coordinates corresponding to the reference positions in adjacent images of all areas covered by the target object, and the field-of-view size information of each image, overlapping areas in adjacent images are determined, thus reducing the time spent finding overlapping areas in adjacent images. Furthermore, in subsequent processes, only the relative offset parameters of the overlapping areas in adjacent images need to be calculated, without needing to calculate all areas of each adjacent image, thereby reducing the computational load. Based on the obtained relative offset parameters, adjacent images can be stitched together to obtain the target image, and thus the target object surface quality data. Therefore, the target object surface quality detection method provided in this specification's embodiments can improve detection efficiency. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this specification, the drawings used in the description of the embodiments of this specification or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 A flowchart of a method for detecting the surface quality of a target object, as illustrated in an embodiment of this specification, is shown.

[0027] Figure 2 This specification illustrates a flowchart of the displacement of a corresponding region of a target object in an embodiment of the present specification.

[0028] Figure 3 This specification illustrates a schematic diagram of target object region division in an embodiment of the present invention.

[0029] Figure 4 A flowchart illustrating the calculation of the relative offset parameter of an overlapping region in adjacent images in an embodiment of this specification is shown.

[0030] Figure 5 This illustrates a schematic diagram of the structure of each image in adjacent images;

[0031] Figure 6 A kind of Figure 5 A schematic diagram of the structure of the target image after stitching together adjacent images;

[0032] Figure 7 This specification shows a schematic diagram of the structure of a target object surface quality detection device according to an embodiment of the present specification;

[0033] Figure 8 Fig. 1 shows a structural schematic diagram of a target object surface quality detection system according to an embodiment of the present specification;

[0034] Figure 9 Fig. 1 shows a structural schematic diagram of a target object surface quality detection system according to an embodiment of the present specification;

[0035] Figure 10 Fig. 1 shows a structural schematic diagram of a target object surface quality detection system according to an embodiment of the present specification; DETAILED DESCRIPTION

[0036] As can be known from the background, with the existing detection scheme, a large amount of time is required to determine the overlapping region between each adjacent image, and the time spent on image stitching is relatively long, thereby resulting in low detection efficiency.

[0037] To solve the above problems, the present specification provides a target object surface quality detection method, which can determine the overlapping region in each adjacent image based on the coordinates corresponding to the reference position in each adjacent image and the field of view size information of each image in the images obtained for all regions covered by the target object, thereby reducing the time for finding the overlapping region of adjacent images, and in the subsequent process, only the relative offset parameters of the overlapping region in each adjacent image need to be calculated, without the need to calculate all regions of each adjacent image, thereby reducing the amount of calculation, and according to the obtained relative offset parameters, each adjacent image can be stitched to obtain a target image, and then the target object surface quality data is obtained. Therefore, the target object surface quality detection method provided by the present specification can improve the detection efficiency.

[0038] In specific implementations, the target object surface quality detection method provided by the present specification can be widely applied to various manufacturing scenarios, such as semiconductor-related manufacturing scenarios, insulator-related manufacturing scenarios, and conductor-related manufacturing scenarios. Accordingly, the target object surface quality detection method provided by the present specification can detect the surface quality of various target objects, such as the surface quality (e.g., flatness) of a wafer. The present specification does not specifically limit the application scenarios of the target object surface quality detection method and the types of target objects to be detected.

[0039] To enable those skilled in the art to better understand and implement the embodiments of the present specification, the following will be described in detail with reference to the accompanying drawings and through specific application examples.

[0040] Reference Figure 1 Fig. 1 shows a structural schematic diagram of a target object surface quality detection system according to an embodiment of the present specification; Figure 1 As shown in Fig. 1, in some embodiments of the present specification, the surface quality of a target object can be detected according to the following steps.

[0041] S11, acquire a plurality of images of the target object.

[0042] The acquisition process comprises: moving the respective regions of the target object to a preset reference position respectively, and acquiring images of the target object at the reference position to obtain images of the target object covering all regions.

[0043] Specifically, in the process of acquiring the target image, the target object can be regionally divided, and the target object can be controlled to move, and when it is determined that the corresponding region is located at the preset reference position, the image corresponding to the region can be acquired, and so on until the images corresponding to all regions of the target image are acquired.

[0044] In a specific implementation, the surface of the target object can be divided based on the shape of the target object to obtain a plurality of regions for representing the surface of the target object. For example, if the target object is a square structure, the plurality of regions obtained by division can all be square. The embodiments of the present specification do not limit the target object and the respective regions of the target object, as long as the respective regions of the target object can cover the surface of the target object.

[0045] S12, determine the overlapping region in each adjacent image according to the coordinates corresponding to the reference position in each adjacent image of the images of all regions covered by the target object and the field of view size information of each image.

[0046] Specifically, by moving the target object, the respective regions of the target object can be moved to the preset reference position respectively, and then the image corresponding to each region of the target object acquired has an exact coordinate value, and the overlapping region in each adjacent image is determined according to the coordinates corresponding to the reference position in each adjacent image and the field of view size information of each image.

[0047] In an optional example, the overlapping region in each adjacent image can be determined according to the difference between the coordinates corresponding to the reference position in each adjacent image and the field of view size information of each image.

[0048] It should be noted that in a specific implementation, based on the size of the target object, corresponding field of view size information can be selected to acquire the image corresponding to the target object. For example, if the target image is large, a larger field of view size information can be selected to sample each region of the target image; if the target image is small, a smaller field of view size information can be selected to sample each region of the target image. And in the same detection process, the field of view size information of each image can be the same.

[0049] S13, calculate the relative offset parameters of the overlapping region in each adjacent image respectively.

[0050] Specifically, in the actual detection process, in order to obtain the image corresponding to the target object, the relative offset parameters of the overlapping regions in each adjacent image can be calculated, and the subsequent process can splice the overlapping regions in each adjacent image according to the calculated relative offset parameters.

[0051] S14, according to the relative offset parameters, splicing each adjacent image to obtain a target image.

[0052] Specifically, the calculated relative offset parameters can be used to represent the relative pose offset between each adjacent image (for example, the offset of each adjacent image in the longitudinal and transverse directions), and according to the relative offset parameters, each adjacent image can be subjected to displacement processing, and the overlapping regions in each adjacent image can be spliced, so that the target image for representing the entire surface of the target object can be obtained.

[0053] S15, detecting the target image to obtain the surface quality data of the target object.

[0054] Specifically, through the above steps S11 to S14, the target image for representing the entire surface of the target object can be obtained, and by detecting the corresponding parameters of the target image, the surface quality data of the target object can be obtained.

[0055] As a specific example, if the target object is a wafer, the flatness value of the wafer image surface is detected to determine whether the wafer surface is flat.

[0056] Therefore, by using the above detection method, since the overlapping regions in each adjacent image can be determined, the time for finding the overlapping regions of adjacent images can be reduced, and in the subsequent process, only the relative offset parameters of the overlapping regions in each adjacent image need to be calculated, without the need to calculate all regions of each adjacent image, thereby reducing the amount of calculation. Therefore, the detection efficiency can be improved.

[0057] In order for those skilled in the art to better understand and implement the embodiments of the present specification, the concepts, schemes, principles and advantages of the embodiments of the present specification are described in detail below with reference to the drawings and through specific application examples.

[0058] As described above, in the process of obtaining a plurality of images of the target object, the target object can be moved multiple times to ensure that the target object is within the sampling field of view, and the corresponding image is obtained when the corresponding region of the target object is determined to be at the reference position. That is, through the above displacement control process, the target object can always be within the sampling field of view.

[0059] In some embodiments of the present disclosure, the change of the target object relative to the reference position can be characterized by the coordinate change of the motion platform. Specifically, the corresponding regions of the target object can be moved to the reference position by controlling the motion platform to move the target object, and the image acquisition device can be controlled to acquire images of the corresponding regions of the target object at the reference position, wherein the target object is placed on the motion platform, and the image acquisition device is arranged above the motion platform.

[0060] Specifically, since the target object is placed on the motion platform, the target object can be moved towards the reference position by controlling the motion of the motion platform. When the target object is determined to be at the reference position, the image acquisition device arranged above the motion platform can acquire images of the corresponding regions of the target object at the reference position. Moreover, by controlling the movement of the motion platform, the corresponding regions of the target object can be moved to the reference position without directly contacting the target object, thereby reducing the probability of damage to the surface of the target object and improving the accuracy of the detection result.

[0061] It can be understood that in some scenarios with relatively low detection accuracy requirements, the corresponding regions of the target object can also be moved to the reference position by directly moving the target object.

[0062] In specific implementations, in the manner of moving the corresponding regions of the target object to the reference position by controlling the motion platform to move the target object, the corresponding regions of the target object can be moved to the reference position based on the coordinates before and after the motion of the motion platform.

[0063] As a specific example, referring to the displacement flowchart of a corresponding region of a target object in an embodiment of the present disclosure shown in FIG. 8, wherein the corresponding region of the target object is provided with a marker point, as shown in FIG. 9, the corresponding region of the target object can be moved to the reference position according to the following steps. Figure 2 Figure 2 S21, the motion platform is controlled to move, and the image acquisition device is controlled to acquire first and second images containing the marker point in the corresponding region of the target object, and the coordinate position of the marker point in the first and second images relative to the motion platform is calculated according to a pre-acquired template image containing the marker point.

[0064] S21, the motion platform is controlled to move, and the image acquisition device is controlled to acquire first and second images containing the marker point in the corresponding region of the target object, and the coordinate position of the marker point in the first and second images relative to the motion platform is calculated according to a pre-acquired template image containing the marker point.

[0065] ​Specifically, since the target object is placed on a motion platform, by controlling the movement of the motion platform, the marker points of the corresponding area of ​​the target object can be placed within the field of view of the image acquisition device. At this time, a first image containing the marker points can be acquired, and based on a pre-acquired template image containing the marker points, the coordinate position of the marker points in the first image relative to the motion platform is obtained through an alignment algorithm. Afterwards, the motion platform can be moved again, so that the corresponding area is again within the field of view of the image acquisition device. At this time, a second image containing the marker points can be acquired, and based on the pre-acquired template image, the coordinate position of the marker points in the second image relative to the motion platform is calculated.

[0066] In some embodiments of this specification, the template image may be created based on image features of the target object. For example, it may be created based on factors such as the size of the target object.

[0067] S22, calculate the first offset between the coordinate position of the marker point in the first image and the reference position based on the pixel parameters of the image acquisition device.

[0068] Specifically, based on the coordinates of the marker points in the first image and the pixel parameters of the image acquisition device, the first offset between the marker points in the first image and the reference position can be obtained.

[0069] In some embodiments of this specification, the pixel parameters of the image acquisition device may include pixel equivalents, wherein the pixel equivalents represent the actual physical size represented by one pixel in the acquired image.

[0070] For example, if the coordinate position of the marker point of the first image relative to the motion table is (x... 11 y 11 If the pixel equivalent of the image acquisition device is u, then the first offset (x1 = u * x) 11 y1=u*y 11 ).

[0071] S23, based on the first offset, control the motion stage to move so that the marker point of the first image coincides with the reference position, and record the first coordinate of the motion stage.

[0072] Specifically, based on the calculated first offset, the motion stage is controlled to move by a distance corresponding to the first offset, so that the marked point in the first image coincides with the reference position, and the first coordinates of the motion stage are recorded.

[0073] For example, if the first offset is (x1, y1), the motion stage is controlled to move x1 in the x direction first, and then move y1 in the y direction; or the motion stage is controlled to move y1 in the y direction first, and then move x1 in the x direction. The embodiments of the present application do not limit the motion sequence of the motion stage, as long as the target object can be moved to the reference position according to the first offset.

[0074] S24, a second offset between the coordinate position of the mark point in the second image and the reference position is calculated according to the pixel parameters of the image acquisition device.

[0075] Specifically, the second offset between the mark point in the second image and the reference position can be obtained according to the coordinate position of the mark point in the second image and the pixel parameters of the image acquisition device.

[0076] For example, if the coordinate position of the mark point in the second image relative to the motion stage is (x 12 , y 12 ), and the pixel equivalent of the image acquisition device is u, the second offset is (x2=u*x 12 , y2=u*y 12 ).

[0077] S25, the motion stage is controlled to move according to the second offset, so that the mark point in the second image coincides with the reference position, and the second coordinate of the motion stage is recorded.

[0078] Specifically, the motion stage is controlled to move according to the second offset, so that the coordinate position of the mark point in the second image coincides with the reference position, and the second coordinate of the motion stage is recorded.

[0079] For example, if the second offset is (x2, y2), the motion stage is controlled to move x2 in the x direction first, and then move y2 in the y direction; or the motion stage is controlled to move y2 in the y direction first, and then move x2 in the x direction. The embodiments of the present application do not limit the motion sequence of the motion stage, as long as the target object can be moved to the reference position according to the second offset.

[0080] S26, a photographing offset angle is obtained according to the first coordinate and the second coordinate of the motion stage.

[0081] Specifically, a pre-set calculation method can be adopted to obtain the corresponding photographing offset angle according to the first coordinate and the second coordinate of the motion stage.

[0082] As an optional example, assuming that the first coordinate of the motion table is (X1, Y1) and the second coordinate is (X2, Y2), the photographing offset angle angle = arctan((Y2-Y1) / (X2-X1)).

[0083] It should be noted that in the embodiments of the present application, the coordinate change of the motion table can be represented by the coordinates of the same point before and after the motion table moves.

[0084] S27, according to the photographing offset angle, controlling the motion table to move to drive the corresponding regions of the target object to move to the reference positions respectively.

[0085] Specifically, according to the photographing offset angle calculated in step S26, the relationship between the corresponding regions of the target object and the reference positions is adjusted in the opposite direction of the photographing offset angle.

[0086] For example, if the calculated offset angle is 35°, the motion table can be moved in the direction of -35° until the corresponding regions of the target object move to the preset reference positions respectively.

[0087] Therefore, according to the calculated photographing offset angle, the corresponding regions of the target object can be moved to the reference positions, and since the photographing offset angle is obtained by the coordinate positions of the first image and the second image of the corresponding regions of the target object on the motion table and the pixel parameters of the image acquisition device, the alignment of the target object coordinate system and the image acquisition device coordinate system can be realized.

[0088] It should be noted that some steps in the above embodiments do not have a certain order, and can be executed simultaneously or in sequence without contradiction, and the order can be changed. For example, in the actual execution process, step S23 can be executed after step S24, or steps S24 and S25 can be executed first, and then steps S22 and S23 can be executed, or steps S22 and S24 can be executed first, and then steps S23 and S25 can be executed respectively. The embodiments of the present application do not limit the order of steps as long as the corresponding photographing offset angle can be obtained according to the calculated offset.

[0089] In some embodiments of the present disclosure, the coordinate positions of the marker points in the acquired first image and the second image relative to the motion stage can be different from their actual positions, which can affect the accuracy of the subsequently calculated photographing offset angle. Based on this, in a specific implementation, the first image and the second image can be respectively slid through the template image, and a filtering operation can be performed to obtain a matching point with the maximum filtering response value in the first image and the second image respectively, the matching point representing the initial coordinate position of the marker point in the corresponding image relative to the motion stage. Then, interpolation upsampling processing is performed within a preset range of the matching point to obtain a plurality of candidate matching points corresponding to the matching point, and based on the template image, a candidate matching point with the optimal gradient parameter is obtained and taken as the coordinate position of the marker point in the corresponding image relative to the motion stage.

[0090] Specifically, since the template image can contain all feature parameters of the target image, by sliding the first image and the second image through the template window and performing a filtering operation at the same time, a matching point with the maximum filtering response value can be obtained. At this time, interpolation upsampling processing can be performed on the matching point with the maximum response value within a preset range to obtain a plurality of corresponding candidate matching points. A candidate matching point with the maximum gradient point and gradient direction is searched from the plurality of candidate matching points, and the candidate matching point is the coordinate position of the marker point in the image relative to the motion stage. Finally, the coordinate position of the motion stage is obtained according to the coordinate position of the image relative to the motion stage.

[0091] The following takes a circular structure as an example to illustrate in detail how to obtain the coordinate of the motion stage.

[0092] In combination with Figure 2 , referring to a region division schematic diagram of a target object shown in Figure 3 , the target object 30 can contain a plurality of regions (for example, regions A, B, C, and D in the figure), and each region includes a marker point (such as the square structure in regions A, B, C, and D).

[0093] For the convenience of understanding, the image corresponding to the collection region A is taken as an example for illustration.

[0094] Using the method shown in Figure 2 , the first image and the second image containing the marker point in the region A can be collected by moving the motion stage twice, and the coordinate positions of the marker points in the first image and the second image relative to the motion stage can be obtained according to the template image containing the marker point obtained in advance.

[0095] Then, the first image can be subjected to a filtering operation and interpolation upsampling processing to obtain a candidate matching point with the optimal gradient point and gradient direction, and the coordinate position of the marker point in the first image relative to the motion stage (px1 y1 ), according to the pixel equivalent u of the image acquisition device and the width W and the height H of the acquired image, the corresponding photographing point (q x1 y1 ) and the coordinate offset (d x1 y1 ), the first offset of the coordinate position of the mark point in the first image and the reference position is (t x1 y1 ), wherein: d x1 = u (p x1 -0.5W), d y1 = u (p y1 -0.5H), t x1 =p x1 +d x1 , t y1 =p y1 +d y1 . Accordingly, the second offset of the coordinate position of the mark point in the second image and the reference position (t x2 y2 ) can be obtained.

[0096] Then, according to the first offset, the motion stage is controlled to move until the mark point of the first image coincides with the reference position, and the first coordinate of the motion stage is obtained; and according to the second offset, the motion stage is controlled to move until the mark point of the second image coincides with the reference position, and the second coordinate of the motion stage is obtained.

[0097] It should be noted that the shape of the target object and the division of each region, and the shape of the mark point on each region are only examples for illustration, and the present specification does not limit the specific shape of the target object and the shape of the mark point on each region of the target object.

[0098] Through the above process, the corresponding regions of the target object can be moved to the reference position respectively, and the images corresponding to the corresponding regions can be obtained. Then, the relative offset parameters of the overlapping regions in each adjacent image can be calculated, and each adjacent image can be spliced according to the obtained offset parameters to obtain a target image for representing the surface of the target object.

[0099] In a specific implementation, referring to a calculation flowchart of the relative offset parameters of the overlapping regions in adjacent images as shown in FIG. 6, the calculation can be performed according to the following steps. Figure 4 Figure 4 S41, based on the first coordinate and the second coordinate of the motion stage corresponding to each region located at the reference position, the coordinate offset value of the motion stage is obtained.

[0100] S41, based on the first coordinate and the second coordinate of the motion stage corresponding to each region located at the reference position, the coordinate offset value of the motion stage is obtained. ​​​​​​

[0101] Specifically, as known from the foregoing, the first coordinate and the second coordinate of the motion table can be obtained by acquiring the image of the target object with the corresponding region of the target object located at the reference position, and the coordinate offset value before and after the motion table moves can be obtained according to the first coordinate and the second coordinate.

[0102] In an optional example, the coordinate offset value before and after the motion table moves can be obtained based on the difference between the first coordinate and the second coordinate.

[0103] S42, obtaining a pixel offset corresponding to each adjacent image according to the coordinate offset value of the motion table and the pixel parameter of the image acquisition device, as the blur position of the overlapping region in each adjacent image.

[0104] Specifically, the pixel offset corresponding to each adjacent image obtained according to the coordinate offset value of the motion table and the pixel parameter of the image acquisition device cannot accurately represent the offset between each image and the reference position, and the obtained pixel offset only represents the blur position of the overlapping region in each adjacent image.

[0105] S43, selecting a region with a preset width as the candidate overlapping region of each adjacent image at a preset magnification with the blur position of the overlapping region in each adjacent image as the center.

[0106] Specifically, in actual operation, each region of the target object corresponds to an image with a blur position. The inventors have found in a large number of experiments that the true position of the overlapping region in each adjacent image fluctuates within a certain range of the blur position at a preset magnification, and therefore a region with a preset width can be cropped with the blur position of the overlapping region in each adjacent image as the center, as the candidate overlapping region of each adjacent image.

[0107] As a specific example, a 100-pixel by 100-pixel region centered on the candidate overlapping region of each adjacent image can be selected as the candidate overlapping region of each adjacent image at a 5x magnification.

[0108] S44, matching the candidate overlapping region of each adjacent image with the overlapping region of each adjacent image respectively to obtain a point with the largest gradient change corresponding to each image in each adjacent image.

[0109] Specifically, the candidate overlapping region of each adjacent image includes a plurality of points, and by matching the candidate overlapping region of each adjacent image with the overlapping region of each adjacent image respectively, a point with the largest gradient change can be obtained, which can be used to represent the true position of the overlapping region in each adjacent image.

[0110] S45, calculating the coordinate difference of the points with the maximum gradient change in each adjacent image to obtain the offset parameter of the overlapping region in each adjacent image.

[0111] Specifically, each image in the adjacent images contains a point with the maximum gradient change in the overlapping part of the adjacent images, and the offset parameter of the overlapping region in each adjacent image can be obtained according to the coordinate difference of the points with the maximum gradient change in the two images in the adjacent images.

[0112] For example, if the coordinates of the point with the maximum gradient change in one of the adjacent images are (t Y1 , t X2 ), and the coordinates of the point with the maximum gradient change in the other image are (t Y2 , t x ), then the coordinates of the offset parameter (off X2 = t X1 -t y , off Y2 = t Y1 -t x1 ) are obtained.

[0113] In a specific implementation, when a region with a preset width is selected as a candidate overlapping region of each adjacent image at the center of the blurred position of the overlapping region in each adjacent image, the points on the candidate overlapping region may not meet the requirements, and in this case, appropriate points need to be found as the true positions of the overlapping region in each adjacent image. Therefore, before the calculation of the coordinate difference of the points with the maximum gradient change in each adjacent image to obtain the offset parameter of the overlapping region in each adjacent image, the method further includes: calculating the derivative value of the gray value of the edge of the overlapping region in each adjacent image and the convolution sum of the overlapping region in each adjacent image to obtain the corresponding response value; calculating the derivative value of the gray value of the edge of the candidate overlapping region of each adjacent image and calculating the convolution value of the overlapping region in each adjacent image, and superimposing the two to obtain the response value of the candidate region of each adjacent image; comparing the response value of the candidate region of each adjacent image with the response value of the image of the overlapping region of each adjacent image to obtain the matching parameter corresponding to each adjacent image; and if the matching parameter is greater than a preset threshold, calculating the offset parameter of the overlapping region in each adjacent image.

[0114] Specifically, when the convolution calculation is performed, the convolution value of the overlapping region and the overlapping region in each adjacent image is the largest, and correspondingly, the sum of the derivative value of the gray value of the edge of the overlapping region in each adjacent image is also the largest, and the obtained response value is also the largest. In order to facilitate calculation, the response value can be normalized, that is, the maximum response value is 1.

[0115] Then, the derivative values of the edge gray values of the candidate overlapping regions of each adjacent image are calculated respectively, and the convolution sum of each adjacent image is obtained, so as to obtain the response values of the candidate regions of each adjacent image.

[0116] The response values of the candidate regions of each adjacent image are compared with the maximum response value, for example, the response values of the candidate regions of each adjacent image are divided by the maximum response value to obtain the corresponding matching parameters. Then, the matching parameters are compared with the preset matching threshold, and when the matching parameters are greater than the preset matching threshold, the offset parameters of the overlapping regions of each adjacent image are calculated; otherwise, the appropriate points are re-found as the real positions of the overlapping regions of each adjacent image.

[0117] By using the scheme in the above embodiment, the relative offset parameters of the overlapping regions of each adjacent image are calculated respectively, and then the images can be spliced according to the relative offset parameters.

[0118] In a specific implementation, the offset parameters can be used to perform a shift and rotation operation on each adjacent image to obtain a pose-corrected image of each adjacent image, and the pose-corrected images of each adjacent image are spliced to obtain the target image.

[0119] Specifically, the offset parameters calculated by the foregoing embodiment can represent the offset amounts of the overlapping regions of each adjacent image in the x direction and the y direction, and the relative rotation angle, according to the above parameters, each adjacent image can be shifted to realize the alignment of the overlapping regions of each adjacent image, and the aligned images are spliced to obtain the target image.

[0120] For example, Figure 5 The images S1 and S2 corresponding to two adjacent regions of a target object are shown, wherein the images S1 and S2 have the same part, that is, the region F1 (the region shown by the shadow in the image S1) in the image S1 and the region F2 (the region shown by the shadow in the image S2) in the image S2, wherein the region F1 and the region F2 are the same, by calculating the relative offset parameters between F1 and F2, and moving any one of the images according to the relative offset parameters, the splicing of the images S1 and S2 can be realized to obtain a part of the target image S (as shown in the image S). Figure 6 The region F represents the overlapping region after the splicing of the images S1 and S2.

[0121] It should be noted that the shape of the image is only an example for illustration. The embodiments of the present application do not limit the shape of the obtained target corresponding image.

[0122] In the actual detection process, the lens of the image sampling device can be deformed, resulting in a difference between the coordinate values of the images of each region obtained on the motion table and the actual coordinate values, and further causing the non-overlapping regions of adjacent images to be spliced, or the overlapping parts of adjacent images to not be completely spliced, and further causing the detection result to be inaccurate.

[0123] Based on this, in specific implementation, after the image acquisition device is used to collect images of each region of the target object located at the reference position, the image collected by the image acquisition device can also be deformed according to the pre-set deformed matrix.

[0124] Specifically, the error caused by the image acquisition device can be considered in advance, and when the corresponding image is collected, the image can be corrected by using the pre-set deformed matrix.

[0125] In some embodiments of the present specification, the deformed matrix can be obtained in the following manner:

[0126] The image acquisition device is used to collect images of each point on the pre-set calibration board, and the point coordinates corresponding to each point on the calibration board in the images are obtained; according to the point coordinates corresponding to each point on the calibration board and the actual point coordinates of each point on the calibration board obtained in advance, and through a pre-set deformed error model, a corresponding deformed matrix is obtained.

[0127] Specifically, the actual coordinate values of each point on the calibration board can be pre-set, and then the image corresponding to the calibration board is collected by the image acquisition device, and then the image point coordinates corresponding to each point on the calibration board can be obtained. The two are input into the pre-set deformed error model to obtain each parameter in the deformed matrix.

[0128] As a specific example, it is assumed that the coordinates of each point in the calibration board are (b x1 , b y1 , b z ), (b x2 , b y2 , b z ), …, (b xi , b yi , b z ), …, (b xi , b yi , b z ), and the image coordinates corresponding to each point are (a x1 , a y1 ), (a x2 , a y2 ), …, (a xi , a yi), …, (a xn , a yn ), wherein bz represents the thickness of the calibration plate, i and n are integers greater than or equal to 1, and i is less than or equal to n, and there are:

[0129]

[0130] The above coordinate values are sequentially brought into formula (1) to obtain the specific values of f1, β, θ, f2, c x and c y , and then obtain the de-distortion matrix. The de-distortion matrix can be used to correct the coordinate values corresponding to each image, thereby improving the detection accuracy.

[0131] In a specific implementation, in order to more accurately extract edge information representing the structure of an image, the overlapping region of each adjacent image can be filtered after the overlapping region of each adjacent image is determined and before the relative offset parameters of the overlapping region of each adjacent image are calculated.

[0132] Specifically, an image within a preset range from the overlapping region of each adjacent image can be selected, and each point in the image within the preset range can be subjected to bilateral filtering. For example, as shown in FIG. 1B, the images S1 and S2 corresponding to two adjacent regions have an overlapping region F1 and an overlapping region F2. An image within a preset range from the overlapping region F1 can be selected, and an image within a preset range from the overlapping region F2 can be selected, and the two images within the preset ranges can be subjected to filtering to better preserve the edges of the overlapping regions. Figure 4

[0133] In a specific implementation, a multi-stage iterative bilateral filtering method can be used to process the overlapping regions of each adjacent image. Taking a point in an image as an example, the pixel value I of the point subjected to bilateral filtering once is:

[0134]

[0135] wherein p (x，y) represents the coordinates of the point to be filtered, q (x，y) represents the coordinates of the point after filtering, N is the size of the filtering kernel, near (x，y) is the grayscale proximity weight, sim (x，y) is the grayscale similarity weight, ns is near (x，y) and sim (x，y) In the filtering kernel, the sum of the products is δ, which is the Gaussian smoothing factor of the grayscale proximity weight, and ε is the Gaussian smoothing factor of the grayscale similarity weight.

[0136] wherein:

[0137] ​

[0138]

[0139]

[0140] In a specific implementation, a corresponding number of iterations can be set to reduce the filtering time on the basis of preserving the edges. In actual operation, the inventor found that after 5-8 iterations of filtering, the edges of the overlapping area can be well preserved.

[0141] It can be understood that the above-mentioned number of iterations is only an example. Based on different target objects and the accuracy of the image acquisition device used, the number of iterations can be flexibly set until the edges of the overlapping area after filtering reach the preset requirements.

[0142] As mentioned above, the embodiments of the present specification are used to splice the plurality of images of the target object obtained to obtain the target image for representing the surface of the target object, and therefore, the splicing quality between the adjacent images directly determines the quality of the surface of the target object in the subsequent detection process. Based on this, in some embodiments of the present specification, before the target image is detected to obtain the surface quality data of the target object, it can further include: determining whether the target image meets the preset splicing quality requirement.

[0143] In a specific implementation, the following method can be used to determine whether the target image meets the preset splicing quality requirement: performing a subtraction operation on the overlapping area in the target image and each adjacent image before splicing; calculating and comparing the gray value range of the corresponding overlapping area after the subtraction operation, respectively, if the gray value range is within the preset range, it is determined that the gray value range of the corresponding overlapping area in the target image and each adjacent image is consistent.

[0144] Specifically, in combination with reference Figure 5 and Figure 6 The splicing area F of the target image S obtained by splicing is subtracted from the area F1 in the image S1 before splicing and the area F2 in the image S2, respectively, and then the gray distribution of each point in the image F-F1 and F-F2 after analyzing and calculating is analyzed and calculated, respectively. If the image S1 and the image S2 are completely spliced, i.e. the image after splicing has no overlap and no gap, the gray value should be close to 0, and the closer to 0, the better the splicing effect of the adjacent images, and at this time it can be determined that the gray value range of the corresponding overlapping area in the target image and each adjacent image is consistent.

[0145] In actual operation, the target image can not meet the preset splicing quality requirement. The inventors analyze the cause of this situation and find that the related parameters of the filtering process can be too large or too small in the filtering process on the overlapping regions of the adjacent images after the distortion removal, thereby causing the target image to not meet the preset splicing quality requirement.

[0146] In some embodiments of the present disclosure, when it is determined that the target image does not meet the preset splicing quality requirement, the related parameters in the filtering process can be changed, and the iterative filtering process on the overlapping regions of the adjacent images can be performed until the preset iteration stopping condition is met.

[0147] For example, one or more of the following parameters can be changed: the filter kernel size N in the filtering process, or the Gaussian smoothing factor δ of the gray level neighborhood weight.

[0148] In specific implementation, to reduce the detection time, when the iterative filtering process on the overlapping regions of the adjacent images meets the preset iteration stopping condition, the iterative filtering process on the overlapping regions of the adjacent images is stopped.

[0149] In some embodiments of the present disclosure, the preset iteration stopping condition includes at least one of the following:

[0150] The gray level value range of the target image and the corresponding overlapping regions in the adjacent images is consistent, that is, when a certain filtering process is performed, the gray level value obtained by performing the subtraction operation on the target image and the overlapping regions in the adjacent images before splicing is close to 0, and the iterative filtering process on the overlapping regions of the adjacent images is stopped.

[0151] The iteration number reaches a preset iteration number threshold, that is, when the iteration number of the iterative filtering process on the overlapping regions of the adjacent images reaches the threshold, the gray level value obtained by performing the subtraction operation is still not close to 0, and at this time, the iterative filtering process on the overlapping regions of the adjacent images is stopped.

[0152] Further, in some embodiments of the present disclosure, when it is determined that the target image still does not meet the preset splicing quality requirement when the iteration number reaches the preset iteration number threshold, in the subsequent processing process, the filtering process on the overlapping regions of the adjacent images after the distortion removal can be performed by using the filtering parameters of the last time of the preset iteration number threshold.

[0153] For example, after the overlapping regions of the adjacent images are filtered for 5 times, the obtained target image still does not meet the preset splicing quality requirement, and at this time, the relative offset parameters obtained after the fourth filtering process can be used to splice the images.

[0154] Thus, by the above steps, the images of the corresponding regions of the target object located at the preset reference positions can be collected, and the overlapping regions of the adjacent images can be spliced and the splicing quality can be evaluated, so that the target image obtained by splicing meets the preset splicing quality requirement.

[0155] In the actual detection process, due to the incomplete symmetry of the optical path of the image acquisition system itself or the offset of the symmetry point, etc., the gray difference of the splicing edge region of the target image obtained by splicing may be caused, and thus in the specific implementation, the fusion processing can be performed on the splicing edge of the adjacent images after splicing.

[0156] Specifically, based on the distance of each point in the overlapping region of the adjacent images for splicing to the splicing edge, the gray value of each point on each image can be multiplied by different weights, wherein the gray value of the point with a distance less than a preset distance threshold is multiplied by a first weight, the gray value of the point with a distance greater than or equal to the distance threshold is multiplied by a second weight, and the first weight is greater than the second weight.

[0157] As a specific example, the weight value at different distances of the splicing edge can be calculated according to formula (3):

[0158]

[0159] Wherein y represents the value of y direction at the splicing edge (x, y) of the image after splicing, offsety represents the offset of each point on the image to the splicing edge in the vertical direction, and D represents the width of the overlapping region.

[0160] Thus, by multiplying the gray values of the points at different offsets from the image splicing edge by different weights, the gray value difference of different regions of the image can be reduced.

[0161] In the embodiments of the present specification, the gray value size corresponding to the splicing edge can be obtained based on the actual gray value of each point.

[0162] As a specific example, assuming S (x，y) represents the gray value at the splicing edge (x, y) of the image after splicing, S1 (x，y) and S2 (x，y) represent the gray values at the corresponding positions of the two images after splicing, (offsetx, offsety) represent the offsets of each point to the splicing edge in the horizontal and vertical directions respectively, and D is the width of the overlapping region, then after the image fusion processing, the gray value S (x，y) of the splicing edge is:

[0163]

[0164] The embodiment of the present specification also provides a target object surface quality detection device corresponding to the above target object surface quality detection method, which is described in detail below with reference to the drawings through specific examples.

[0165] Referring to Figure 7 The structure schematic diagram of a target object surface quality detection device in the embodiment of the present specification is shown in FIG. 7. In some embodiments of the present specification, the target object surface quality detection device 70 can include:

[0166] An image acquisition unit 71 adapted to acquire a plurality of images of a target object;

[0167] The image acquisition unit 71 is adapted to move the corresponding regions of the target object to the preset reference positions respectively, acquire the images of the target object at the reference positions, and obtain the images of the target object covering all regions.

[0168] An image determination unit 72 adapted to determine the overlapping regions in each adjacent image according to the coordinates corresponding to the preset positions in each adjacent image in the images of the target object covering all regions and the field of view size information of each image.

[0169] A parameter calculation unit 73 adapted to calculate the relative offset parameters of the overlapping regions in each adjacent image respectively.

[0170] An image splicing unit 74 adapted to splice each adjacent image according to the relative offset parameters to obtain a target image.

[0171] An image detection unit 75 adapted to detect the target image to obtain the target object surface quality data.

[0172] By using the above detection device 70, the image acquisition unit 71 can acquire the images of the target object at the reference positions by moving the corresponding regions of the target object to the preset reference positions respectively, thereby obtaining the images of the target object covering all regions. The image determination unit 72 can determine the overlapping regions in each adjacent image based on the coordinates corresponding to the reference positions in each adjacent image in the images of the target object covering all regions and the field of view size information of each image, thereby reducing the time for finding the overlapping regions of adjacent images. In the subsequent process, the parameter calculation unit 73 can only calculate the relative offset parameters of the overlapping regions in each adjacent image, without calculating all regions of each adjacent image, thereby reducing the amount of calculation. According to the obtained relative offset parameters, the image splicing unit 74 can splice each adjacent image to obtain a target image, and the image detection unit 75 can detect the target image to obtain the target object surface quality data. Therefore, the above detection device 70 provided by the embodiment of the present specification can improve the detection efficiency.

[0173] In specific implementations, the image acquisition unit 71 can control the control motion table to move to bring the corresponding region of the target object to the reference position, and control the image acquisition device to acquire the image of the target object at the reference position. When the image acquisition device acquires the image of the corresponding region, the image acquisition unit 71 can acquire the image.

[0174] The image determination unit 72 can determine the overlapping region in each adjacent image according to the coordinate difference of the reference position in each adjacent image of the image of all regions covered by the target object, and the field of view size information of each image.

[0175] In specific implementations, the lens of the image sampling device can be deformed, resulting in a difference between the coordinate value of the obtained image of each region and the actual coordinate value, and further causing the non-overlapping region of the adjacent images to be spliced, or the overlapping part of the adjacent images to be not completely spliced, and a gap exists between the two, resulting in inaccurate detection results. Continuing to refer to Figure 7 , the detection device 70 can further include an image correction unit 76 adapted to perform de-distortion processing on the image of each region of the target object at the reference position acquired by the image acquisition device according to a pre-set de-distortion matrix. The calculation process of the de-distortion matrix can be referred to the description in the foregoing embodiments, which will not be described here.

[0176] In specific implementations, in order to more accurately extract edge information that can represent the structure of the image, continuing to refer to Figure 7 , the detection device 70 can further include an image enhancement unit 77 adapted to perform filtering processing on the overlapping region of each adjacent image after the de-distortion processing. The working principle and process of the image enhancement unit 77 can be referred to the foregoing embodiments, which will not be described here.

[0177] Then, the parameter calculation unit 73 can calculate the offset parameter of the overlapping region in each adjacent image, and the image splicing unit 74 can splice each adjacent image according to the relative offset parameter to obtain the target image. The calculation process of the offset parameter of the overlapping region in each adjacent image and the splicing process of each adjacent image can be referred to the foregoing embodiments, which will not be described here.

[0178] As described above, the embodiments of the present specification are to splice the plurality of images of the target object obtained to obtain the target image representing the surface of the target object, and therefore, the splicing quality between each adjacent image directly determines the quality of the surface of the target object obtained in the subsequent detection process. Based on this, in some embodiments of the present specification, continuing to refer to Figure 7The detection device 70 can further comprise an evaluation unit 78 adapted to determine whether the target image meets preset splicing quality requirements.

[0179] Thus, by the above-mentioned units, the images of the corresponding regions of the target object located at the preset reference positions can be acquired, and the overlapping regions of the adjacent images can be spliced and evaluated for splicing quality, so that the target image obtained by splicing meets the preset splicing quality requirements.

[0180] In actual detection, due to the incomplete symmetry of the optical path of the image acquisition system or the offset of the symmetry point, etc., the gray difference of the splicing edge region of the target image obtained by splicing can be caused. Therefore, continuing to refer to Figure 7 The detection device 70 can further comprise an image fusion unit 79 adapted to perform fusion processing on the splicing edge of the adjacent images after splicing, and the image detection unit 75 is adapted to detect the target image after fusion processing.

[0181] Correspondingly, the present specification also provides a target object surface quality detection system, which will be described in detail below by specific examples with reference to the accompanying drawings.

[0182] Referring to the structure schematic diagram of a target object surface quality detection system in the present specification shown in Figure 8 In some embodiments of the present specification, the target object surface quality detection system 80 can comprise:

[0183] An image acquisition device 81 adapted to acquire the image of the target object;

[0184] The target object surface quality detection device 82 described in the foregoing embodiments is adapted to detect the image of the target object.

[0185] The working principle and specific structure of the target object surface quality detection device can be referred to the foregoing embodiments, which will not be described here.

[0186] In specific implementation, in order to acquire the image of the target object of the corresponding region located at the reference position, and then obtain the image of the target object covering all regions, continuing to refer to Figure 8 The target object surface quality detection system 80 can further comprise a motion stage 83, and the target object is placed on the motion stage, and the image acquisition device 81 is arranged above the motion stage 83.

[0187] Correspondingly, the target object surface quality detection device 82 is further adapted to control the motion stage 83 to move to drive the corresponding region of the target object to move to the preset position, and control the image acquisition device 81 to acquire the image of each region of the target object located at the reference position.

[0188] To better understand and explain the working principle of the target object surface quality detection system in the embodiments of this specification, the following specific examples will be used for illustration.

[0189] Reference Figure 9 The diagram shown here illustrates the structure of a target object surface quality detection system in a specific application scenario of this embodiment. Figure 9 As shown, the target object 9A (e.g., a wafer) is placed on the surface of the motion stage 91, and an image acquisition device 92 (e.g., a camera) for acquiring the target object 9A is disposed above it.

[0190] When it is necessary to acquire an image of the target object 9A, the target object surface quality detection device ( Figure 9 (Not shown) By moving the position of the motion stage 91 twice, the image acquisition device 92 can acquire a first image and a second image of the target object 9A when it is in a reference position, respectively. The reference position can be the center of the sampling field of view of the image acquisition device 92, which can be determined according to a pre-set coordinate system (e.g., the x-axis and y-axis in the figure). Based on the pre-acquired template image and its own pixel parameters, the first offset between the coordinate position of the marker point in the first image and the reference position, and the second offset between the coordinate position of the marker point in the second image and the reference position can be calculated. Then, the target object surface quality detection device can drive the motion stage 91 to move according to the first offset and the second offset. The first coordinates of the motion stage 91 are obtained by aligning the coordinates of the marked point in the first image with the reference position. The second coordinates of the motion stage 91 are obtained by aligning the coordinates of the marked point in the second image with the reference position. Based on the first and second coordinates of the motion stage 91, an offset angle can be captured. Based on the offset angle, the corresponding area of ​​the target object 9A can be moved to the reference position, and a correspondence between the target object coordinate system and the image acquisition device coordinate system is established. The target object surface quality detection device can acquire multiple images of the target object 9A acquired by the image acquisition device 92, and after stitching the multiple images of the target object 9A together, the obtained target image is detected.

[0191] Continue to refer to Figure 9 In practical applications, the target object surface quality inspection system 90 may also include a light source device 93, which is connected to the image acquisition device 92 and located between the image acquisition device 92 and the motion stage 91, and is suitable for providing an illumination environment, such as the light L shown in the figure.

[0192] In practice, the above detection method can be applied to electronic devices, such as... Figure 10The diagram shown is a structural block diagram of the surface quality treatment equipment for the target object provided in the embodiments of this specification. Figure 10 In the above embodiment, the target object surface quality processing device 100 may include a memory 110 and a processor 120. The memory 110 and the processor 120 can communicate with each other through a communication bus 130. The memory 110 stores computer instructions that can be executed on the processor 120. When the processor 120 executes the computer instructions, it can perform the steps of the target object surface quality detection method described in any of the above embodiments. For details, please refer to the above relevant content, which will not be repeated here.

[0193] In specific implementations, the processor may include a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a Field Programmable Gate Array (FPGA), etc. The memory may include Random Access Memory (RAM), Read-Only Memory (ROM), Non-Volatile Memory (NVM), etc.

[0194] In practice, computer instructions may include any suitable type of code implemented using any appropriate high-level, low-level, object-oriented, visual, compiled, and / or interpreted programming language, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc.

[0195] In specific implementation, such as Figure 10 As shown, the target object surface quality processing device 100 may further include a display interface 140 and a display 150 connected via the display interface 140. The display interface 140 can communicate with the memory 110 and the processor 120 via a communication bus 130. The display 150 can display the quality data obtained by the processor 120 executing the target object surface quality detection method provided in the embodiments of this specification.

[0196] This specification also provides a computer-readable storage medium storing computer instructions that, when executed, can perform the steps of the target object surface quality detection method described in any of the above embodiments of this specification. For details, please refer to the above-mentioned related content, which will not be repeated here.

[0197] The computer-readable storage medium can include any suitable type of memory unit, memory device, memory article, memory medium, storage device, storage article, storage medium and / or storage unit. For example, memory, removable or non-removable media, erasable or non-erasable media, writeable or re-writeable media, digital or analog media, hard disk, floppy disk, Compact Disk (CD), DVD, etc. Also, computer instructions can include any suitable type of code, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, etc., implemented using any suitable high-level, low-level, object-oriented, visual, compiled and / or interpreted programming language.

[0198] It should be noted that the terms "first", "second", "third", etc. are used herein only to describe different instances, and cannot be construed to indicate or imply relative importance or imply a quantity of the indicated technical features. Thus, the features defined with the terms "first", "second", "third", etc. can explicitly or implicitly include one or more of the features. Also, the terms "first", "second", "third", etc. are used to distinguish similar objects, and do not necessarily indicate a specific order or represent importance. It can be understood that the terms used in this way can be interchanged as appropriate, so that the embodiments of the present specification described herein can be implemented in an order other than those illustrated or described herein.

[0199] Although the present disclosure is disclosed as above, the present disclosure is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present specification, and therefore the scope of protection of the present disclosure should be defined by the scope defined by the claims.

Claims

1. A method of detecting surface quality of a target object, characterized by, The target object is placed on a moving table, and an image acquisition device is arranged above the moving table. A corresponding region of the target object is provided with a marker point. The method comprises the following steps: A plurality of images of the target object are acquired, including: controlling the moving table to move, so as to control the corresponding region of the target object to move to a preset reference position respectively, and controlling the image acquisition device to acquire images of the target object located at the reference position, thereby obtaining images of the target object covering all regions, According to the coordinates corresponding to the reference position in each adjacent image in the images of the target object covering all regions, and the field of view size information of each image, the overlapping region in each adjacent image is determined; The relative offset parameters of the overlapping region in each adjacent image are calculated respectively, including: based on the first coordinate and the second coordinate of the moving table corresponding to the marker point of each region coinciding with the reference position, the coordinate offset value of the moving table is obtained; according to the coordinate offset value of the moving table and the pixel parameters of the image acquisition device, the pixel offset amount corresponding to each adjacent image is obtained as the blurred position of the overlapping region in each adjacent image; taking the blurred position of the overlapping region in each adjacent image as the center, a region with a preset width is selected as the candidate overlapping region of each adjacent image under a preset magnification; the candidate overlapping region of each adjacent image is matched with the overlapping region of each adjacent image respectively, thereby obtaining the point corresponding to the greatest gradient change of each image in each adjacent image; the coordinate difference value of the point with the greatest gradient change of each adjacent image is calculated, thereby obtaining the offset parameter of the overlapping region in each adjacent image; According to the relative offset parameters, each adjacent image is spliced to obtain a target image; The target image is detected to obtain surface quality data of the target object.

2. The detection method according to claim 1, characterized in that, The control of the moving table to move so as to control the corresponding region of the target object to move to the reference position comprises: controlling the moving table to move, and controlling the image acquisition device to acquire a first image and a second image containing the marker point in the corresponding region of the target object respectively, and calculating the coordinate position of the marker point relative to the moving table in the first image and the second image according to a template image containing the marker point obtained in advance; According to the pixel parameters of the image acquisition device, the first offset amount of the coordinate position of the marker point in the first image from the reference position is calculated, and the second offset amount of the coordinate position of the marker point in the second image from the reference position is calculated; According to the first offset amount, the moving table is controlled to move so that the marker point of the first image coincides with the reference position, and the first coordinate of the moving table is recorded; and according to the second offset amount, the moving table is controlled to move so that the marker point of the second image coincides with the reference position, and the second coordinate of the moving table is recorded; The photographing offset angle is obtained according to the first coordinate and the second coordinate of the moving table; According to the photographing offset angle, the moving table is controlled to move so as to drive the corresponding region of the target object to move to the reference position.

3. The detection method according to claim 2, characterized in that, The step of calculating the coordinate positions of the marker points relative to the motion platform in the first image and the second image based on a pre-acquired template image containing the marker points includes: The template image is used to slide the first image and the second image respectively, and a filtering operation is performed to obtain the matching point with the largest filtering response value in the first image and the second image respectively. The matching point represents the initial coordinate position of the marker point of the corresponding image relative to the motion table. Interpolation upsampling is performed within a preset range of the matching point to obtain multiple candidate matching points corresponding to the matching point; Based on the template image, the candidate matching point with the optimal gradient parameters is obtained, and it is used as the coordinate position of the corresponding image's marker point relative to the motion table.

4. The detection method according to claim 2, characterized in that, The template image is created based on the image features of the target object.

5. The detection method according to claim 1, before calculating the coordinate difference of the point with the largest gradient change in each adjacent image to obtain the offset parameter of the overlapping region in the adjacent images, further includes: Calculate the derivative of the gray value of the edge of the overlapping region in each adjacent image, and calculate the convolution value of the overlapping region in each adjacent image. Then, superimpose the two to obtain the corresponding response value. Calculate the derivative values ​​of the edge gray values ​​of the candidate overlapping regions in each adjacent image, and calculate the convolution values ​​of the overlapping regions in each adjacent image. Then, superimpose the two values ​​to obtain the response values ​​of the candidate regions in each adjacent image. The response values ​​of the candidate regions of each adjacent image are compared with the response values ​​of the overlapping regions of each adjacent image to obtain the matching parameters corresponding to each adjacent image. If the matching parameter is greater than the preset matching threshold, then the offset parameter of the overlapping region in each adjacent image is calculated.

6. The detection method according to claim 1, wherein stitching adjacent images together according to the relative offset parameter to obtain the target image includes: Based on the offset parameter, a shift operation is performed on each adjacent image to obtain a pose correction image of each adjacent image; The pose-corrected images of the adjacent images are stitched together to obtain the target image.

7. The method of claim 1, wherein, After acquiring images of each region of the target object at the reference position using an image acquisition device, the method further includes: Based on a pre-set distortion correction matrix, the images of each region of the target object acquired by the image acquisition device at the reference position are subjected to distortion correction processing.

8. The detection method according to claim 7, characterized in that, The distortion correction matrix is ​​obtained in the following way: The image acquisition device is used to acquire images of each point on a pre-set calibration board, and the coordinates of each point on the calibration board in the image are obtained. Based on the point coordinates corresponding to each point on the calibration board and the pre-acquired actual point coordinates corresponding to each point on the calibration board, and through a preset distortion error model, the corresponding distortion correction matrix is ​​obtained.

9. The detection method according to claim 7, characterized in that, After determining the overlapping regions in each adjacent image, and before calculating the relative offset parameters of the overlapping regions in each adjacent image, the method further includes: The overlapping regions of adjacent images after the distortion correction process are filtered.

10. The surface quality inspection method according to claim 9, wherein The filtering processing on the overlapping area of each adjacent image after the de-distortion processing is performed, comprising: selecting an image within a preset range from the overlapping area of each adjacent image, and performing bilateral filtering processing on each point in the image within the preset range.

11. The method of claim 7, wherein, Before the target image is detected to obtain the target object surface quality data, further comprising: determining that the target image meets the preset splicing quality requirement.

12. The detection method of claim 11, wherein, The determination that the target image meets the preset splicing quality requirement comprises: performing a subtraction operation on the target image and the overlapping area in each adjacent image before splicing; respectively calculating and comparing the gray value range of the corresponding overlapping area after the subtraction operation, and if the gray value range is within a preset range, determining that the gray value range of the corresponding overlapping area in the target image and each adjacent image is consistent.

13. The detection method according to claim 12, characterized in that, Further comprising: when it is determined that the target image does not meet the preset splicing quality requirement, changing the related parameters in the filtering processing process, and iteratively filtering the overlapping area of each adjacent image until the preset iteration stopping condition is met.

14. The detection method according to claim 13, characterized in that, The iterative filtering processing on the overlapping area of each adjacent image until the preset iteration stopping condition is met comprises at least one of the following: the gray value range of the corresponding overlapping area in the target image and each adjacent image is consistent; the number of iterations reaches a preset iteration number threshold.

15. The assay of any one of claims 1-14, wherein, Further comprising: performing fusion processing on the splicing edge of each adjacent image after splicing.

16. The detection method of claim 15, wherein, The fusion processing on the splicing edge of each adjacent image after splicing comprises: based on the distance between each point in the overlapping area of each adjacent image for splicing and the splicing edge, multiplying the gray value of each point on each image by different weights, wherein the gray value of a point with a distance less than a preset distance threshold is multiplied by a first weight, the gray value of a point with a distance greater than or equal to the distance threshold is multiplied by a second weight, and the first weight is greater than the second weight.

17. A target object surface quality detection device, characterized by, Comprising: an image acquisition unit adapted to acquire a plurality of images of a target object, wherein the image acquisition unit is adapted to control a motion platform to move to control a corresponding region of the target object to move to a preset reference position, and to control an image acquisition device to acquire an image of the target object located at the reference position to obtain an image of the target object covering all regions, the image acquisition device being arranged above the motion platform, and the corresponding region of the target object being provided with a marker point; an image determination unit adapted to determine an overlapping area in each adjacent image according to coordinates corresponding to the reference position in each adjacent image in the image of the target object covering all regions and field of view size information of each image; The parameter calculation unit is suitable for calculating the relative offset parameters of the overlapping areas in each adjacent image, including: obtaining the coordinate offset value of the motion table based on the first coordinate and the second coordinate of the motion table when the marking points of each area coincide with the reference position; obtaining the pixel offset of each adjacent image corresponding to the pixel parameters of the image acquisition device as the blur position of the overlapping area in each adjacent image; selecting a region with a preset width as the candidate overlapping area of each adjacent image at a preset magnification ratio with the blur position of the overlapping area in each adjacent image as the center; matching the candidate overlapping area of each adjacent image with the overlapping area of each adjacent image to obtain the point with the largest gradient change of each image in each adjacent image; calculating the coordinate difference value of the point with the largest gradient change of each adjacent image to obtain the offset parameter of the overlapping area in each adjacent image; The image stitching unit is suitable for stitching each adjacent image according to the relative offset parameters to obtain a target image. The image detection unit is suitable for detecting the target image to obtain the surface quality data of the target object.

18. The detection device of claim 17, wherein, Further comprising: The image correction unit is suitable for performing distortion removal processing on the image collected by the image acquisition device with each area of the target object located at the reference position according to a pre-set distortion removal matrix.

19. The detection device of claim 18, wherein, Further comprising: The image enhancement unit is suitable for performing filtering processing on the overlapping area of each adjacent image after the distortion removal processing.

20. The detection device of claim 19, wherein, Further comprising: The evaluation unit is suitable for determining whether the target image meets the preset stitching quality requirement.

21. The detection device according to any one of claims 17-20, characterized in that, Further comprising: The image fusion unit is suitable for performing fusion processing on the stitching edge of each adjacent image after stitching.

22. A target object surface quality detection system, comprising: Including: The image acquisition device is suitable for collecting the image of the target object. The target object surface quality detection device according to any one of claims 17-21 is suitable for detecting the image of the target object.

23. The detection system of claim 22, wherein, Further comprising: The motion table, on which the target object is placed, is arranged above the motion table; The target object surface quality detection device is further suitable for controlling the motion table to move to drive the corresponding area of the target object to move to the reference position, and controlling the image acquisition device to collect the image with each area of the target object located at the reference position.

24. A target object surface quality processing apparatus comprising a memory and a processor, wherein, The memory is suitable for storing one or more computer instructions, and the processor executes the computer instructions to perform the target object surface quality detection method according to any one of claims 1-16.

25. A computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are executed to perform the target object surface quality detection method according to any one of claims 1-16.

Images