A Smart Control Method for Exposure Lamps in an Exposure Machine

Abstract

This invention relates to the field of photographic surface positioning technology, specifically to an intelligent control method for an exposure lamp in an exposure machine. The method includes: acquiring the theoretical position region of a real marking template in a marking image, and several candidate regions in the marking image; selecting several alternative marking regions from all candidate regions based on the difference between each candidate region and the theoretical position region; and acquiring the real marking region based on the boundary symmetry of each alternative marking region and the overlap between the alternative marking regions, for adjusting the position of the exposure lamp. This invention enables accurate adjustment of the exposure lamp position for precise exposure of materials.

Description

Technical Field

[0001] This invention relates to the field of positioning technology on photographic surfaces, and more specifically to an intelligent control method for exposure lamps in an exposure machine. Background Technology

[0002] In the field of exposure equipment, exposure machines typically use mechanical positioning to initially coarsely position the exposure lamp, followed by optical imaging control to identify and precisely align the alignment marks, thereby determining the accurate position of the material and ensuring material precision during the exposure process. In this process, the positioning of the exposure lamp is the core control element of the exposure machine, and its accuracy directly affects the uniformity of exposure and the alignment precision.

[0003] Existing problem: Inaccurate positioning of the exposure lamp can easily cause imaging deviations, leading to mark recognition errors and further affecting the accuracy of material positioning. Especially in practical applications, when there is contamination on the material surface, bright background spots, or reflection and transmission interference during multi-layer stacking, the mark image is prone to ghosting or double images, making it difficult for the system to determine the true mark, resulting in misjudgment and affecting the accurate positioning of the exposure lamp. Summary of the Invention

[0004] This invention provides an intelligent control method for exposure lamps in an exposure machine to solve existing problems.

[0005] The present invention provides an intelligent control method for exposure lamps in an exposure machine, which adopts the following technical solution:

[0006] One embodiment of the present invention provides an intelligent control method for exposure lamps in an exposure machine, the method comprising the following steps:

[0007] Obtain the theoretical location region of the real marker template in the marker image, as well as several candidate regions detected by edge detection in the marker image;

[0008] Based on the difference between each candidate region and the theoretical location region, determine the degree of difference between each candidate region and the template; based on the magnitude of the degree of difference between each candidate region and the template, select several candidate marking regions from all candidate regions;

[0009] Based on the boundary symmetry of each candidate marking region and the overlap between candidate marking regions, determine the probability that each candidate marking region is a real marking region;

[0010] Based on the probability that each candidate marking region is a real marking region, the real marking region is obtained; based on the coordinate position of the real marking region in the marking image, the position of the exposure lamp is adjusted.

[0011] Furthermore, the specific steps for determining the difference between each candidate region and the template are as follows:

[0012] Based on the differences in area and length and width between each candidate region and the theoretical location region, the dissimilarity between each candidate region and the template region is determined.

[0013] Obtain the boundary length of each candidate region, and denote it as the target length;

[0014] Obtain the boundary length of the theoretical location region, and denote it as the standard length;

[0015] When the target length is less than or equal to the standard length, the dissimilarity between the boundary of each candidate region and the template is determined based on the matching similarity of the boundary of each candidate region on the boundary of the theoretical location region.

[0016] When the target length is greater than the standard length, the dissimilarity between each candidate region and the template boundary is determined based on the matching similarity between the boundary of the theoretical location region and the boundary of each candidate region.

[0017] Obtain the normalized value of the boundary dissimilarity between each candidate region and the template, and the mean value of the region dissimilarity, and denote it as the difference between each candidate region and the template.

[0018] Furthermore, the specific steps for determining the regional dissimilarity between each candidate region and the template based on the area and length-width differences between each candidate region and the theoretical location region are as follows:

[0019] Obtain the area and aspect ratio of the theoretical location region, and denote them as the standard area and standard aspect ratio, respectively.

[0020] The normalized value of the absolute difference between the area of ​​each candidate region and the standard area is obtained and denoted as the first difference value. The normalized value of the absolute difference between the aspect ratio of the minimum bounding rectangle of each candidate region and the standard aspect ratio is obtained and denoted as the second difference value. The mean of the first difference value and the second difference value is denoted as the region dissimilarity between each candidate region and the template.

[0021] Furthermore, when the target length is less than or equal to the standard length, the specific steps for determining the boundary dissimilarity between each candidate region and the template based on the matching similarity of the boundary of each candidate region on the boundary of the theoretical location region are as follows:

[0022] At the boundary of each candidate region, starting from any pixel, we traverse the pixel sequence one by one in a clockwise direction to form a pixel sequence, which is called the target sequence.

[0023] On the boundary of the theoretical location region, starting from the i-th pixel, traverse each pixel in a clockwise direction to obtain a pixel sequence of length equal to the target length, which is denoted as the reference sequence of the i-th pixel on the boundary of the theoretical location region.

[0024] The minimum DTW distance among all pixels on the boundary between the target sequence and the theoretical location region is obtained and denoted as the boundary dissimilarity between each candidate region and the template.

[0025] Furthermore, when the target length is greater than the standard length, the dissimilarity between each candidate region and the template boundary is determined based on the matching similarity between the boundary of the theoretical location region and the boundary of each candidate region. The specific steps include the following:

[0026] On the boundary of the theoretical location region, starting from any pixel, traverse the pixel points one by one in a clockwise direction to form a pixel sequence, which is denoted as the new target sequence.

[0027] On the boundary of each candidate region, starting from the j-th pixel, traverse each pixel in a clockwise direction to obtain a pixel sequence of standard length, which is recorded as the new reference sequence of the j-th pixel on the boundary of each candidate region.

[0028] The minimum DTW distance between the new target sequence and the new reference sequence of all pixels on the boundary of each candidate region is obtained and denoted as the boundary dissimilarity between each candidate region and the template.

[0029] Furthermore, the specific steps for selecting several candidate marked regions from all candidate regions are as follows:

[0030] Among all candidate regions, those with differences less than the preset difference threshold are all marked as candidate regions.

[0031] Furthermore, the specific steps involved in determining the probability that each candidate marker region is a true marker are as follows:

[0032] Obtain the center-symmetric boundary of the boundary of each candidate marked region;

[0033] Obtain the normalized value of the number of overlapping pixels between the boundary and the central symmetric boundary of each candidate marking region, and denote it as the target number;

[0034] Obtain the normalized value of the mean gradient value of all pixels on the boundary of each candidate marking region, and denote it as the first mean.

[0035] Obtain the inversely proportional normalized value of the absolute value of the difference between the boundary length of each candidate marked region and the boundary length of the theoretical location region, and denote it as the boundary length consistency.

[0036] The boundary clarity of each candidate marking region is determined based on the number of targets, the first mean, and the consistency of boundary length.

[0037] Based on the overlap between candidate marker areas, determine the regional non-deviation degree of each candidate marker area;

[0038] Obtain the mean of the boundary clarity and the region non-deviation degree of each candidate marking region, and denote it as the probability that each candidate marking region is a true marking region.

[0039] Furthermore, the specific steps for determining the boundary clarity of each candidate marking region based on the target quantity, the first mean, and the consistency of boundary length are as follows:

[0040] The number of targets, the first mean, and the mean of the consistency of boundary length are denoted as the boundary clarity of each candidate marked region.

[0041] Furthermore, the specific steps for determining the regional non-deviation degree of each candidate marker region based on the overlap between candidate marker regions are as follows:

[0042] Obtain the pixel point corresponding to the mean coordinates of the center points of all candidate marking regions, and record it as the estimated center point of the true mark;

[0043] Obtain the distance between the center point of each candidate marker region and the estimated center point of the actual marker, and record it as the center deviation value of each candidate marker region;

[0044] Get the overlapping area of ​​all candidate marked regions;

[0045] The ratio of the area of ​​the overlapping region to the area of ​​each candidate marked region is recorded as the overlap ratio. The inversely proportional normalized value of the center deviation of each candidate marked region and the mean of the overlap ratio are recorded as the regional non-deviation degree of each candidate marked region.

[0046] Furthermore, the specific steps for obtaining the actual marker region are as follows:

[0047] Among all the possibilities that the candidate marking region is the real marking region, the candidate marking region corresponding to the maximum value is recorded as the real marking region.

[0048] The beneficial effects of the technical solution of the present invention are:

[0049] In this embodiment of the invention, the theoretical position region of the actual marker template in the marker image is obtained, along with several candidate regions in the marker image. Based on the difference between each candidate region and the theoretical position region, several alternative marker regions are selected from all candidate regions. By matching the actual marker template, interference information of the material markers is removed, and the ghosting region of the material markers is obtained. Based on the boundary symmetry of each alternative marker region and the overlap between alternative marker regions, the actual marker region is obtained, which is used to adjust the position of the exposure lamp. Thus, based on the edge conditions of each alternative marker region in the ghosting region and the ghosting distribution of the alternative markers, the actual marker is determined, ensuring accurate adjustment of the exposure lamp position and precise exposure of the material. Therefore, this invention eliminates interference and confirms the actual marker before aligning the markers, enabling precise positioning of the exposure lamp for exposure. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the steps of an intelligent control method for an exposure lamp in an exposure machine according to the present invention.

[0052] Figure 2 This is a schematic diagram of the exposure machine structure. Detailed Implementation

[0053] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent control method for an exposure lamp in an exposure machine according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0055] The following description, in conjunction with the accompanying drawings, details a specific scheme for an intelligent control method for exposure lamps in an exposure machine provided by the present invention.

[0056] Please see Figure 1The diagram illustrates a flowchart of a method for intelligent control of an exposure lamp in an exposure machine according to an embodiment of the present invention. The method includes the following steps:

[0057] Step S001: Obtain the theoretical location region of the real marker template in the marker image, as well as several candidate regions detected by edge detection in the marker image.

[0058] In this embodiment, by acquiring the image of the material markings during the precise positioning stage of the exposure machine, the current material marking graphic area is enhanced, interference graphics are removed, and the true markings of the material are determined. Then, the material markings are accurately positioned to determine the position of the exposure lamp and complete the material exposure.

[0059] It should be noted that the specific process of the first exposure when the exposure machine exposes the material is as follows: (1) The material is placed on the PIN by the loading robot, where the PIN is a positioning device used to fix the position of the material and ensure that the subsequent exposure operation can be carried out accurately. (2) The positioning module rises to position the product to the side (mechanical positioning). (3) Vacuum adsorption of material. (4) Positioning module retracts. (5) The CCD camera performs secondary precise positioning of the material (positioning using markers), where the CCD (Charge-Coupled Device) camera is a digital camera that uses a charge-coupled device as an image sensor. (6) The position of the exposure lamp is adjusted according to the CCD camera in the X, Y, and R directions (three-dimensional directions). (7) The exposure lamp is turned on for exposure. Schematic diagram of the exposure machine structure, as shown below. Figure 2 As shown. Figure 2 The system includes: 1. X-direction movement module; 3. Y-direction movement module; 6. R-direction DD motor; 2. CCD; 4. Discharge line; 5. Positioning module, with markings for the feeding and discharging directions. The DD motor (Direct Drive Motor), also known as a direct-drive motor, is a type of motor that directly drives the load without the need for gears, belts, or ball screws.

[0060] Further explanation is needed: In this embodiment, the markings may appear as ghosting or double images when the CCD performs secondary precise positioning of the material. The real markings are then identified. Before identification, it is necessary to obtain the image of the current material markings. The specific steps for obtaining the material marking image are as follows: (1) Exposure lamp preheating and stabilization: After the exposure lamp is started, the light output is monitored by the sensor, and the current and light intensity are adjusted to ensure uniform and stable illumination. (2) Mechanical positioning: The exposure stage completes coarse positioning according to the preset coordinates of the material, and places the material roughly below the exposure light path. (3) Optical imaging and marking acquisition: The CCD camera or optical sensor images the material markings to obtain the image of the markings. (4) Preliminary image processing: The system performs grayscale processing on the acquired image to obtain the grayscale image around the marking. Among them, grayscale processing is a well-known technology, and the specific method is not described here.

[0061] Obtain the theoretical location region of the real marker template in the marker image, as well as several candidate regions in the marker image.

[0062] It should be noted that in this embodiment, the position, size, and shape (rectangular) of the actual marker template are known, and the position and parameters of the CCD camera during image acquisition are also known. Therefore, the theoretically precise location of the template in the captured image can be determined through geometric transformation and image processing techniques. This is a well-known technique, and the specific method will not be described here. However, when there is contamination on the material surface, bright background spots, or reflection and transmission interference during multi-layer stacking, the marker image is prone to ghosting or double images, which may cause a difference between the theoretical location area and the actual template location area in the image. Therefore, it is necessary to identify the actual marker area in the marker image. Specifically, the marker image undergoes image enhancement processing with histogram equalization and noise point filtering processing with morphological opening and closing operations. Then, the Canny edge detection algorithm is used to perform edge detection on the processed marker image to obtain the edge information and gradient value of each pixel. Based on the edge information, a connected component labeling algorithm is used to obtain several connected regions enclosed by the edges as candidate regions. Histogram equalization, morphological opening and closing operations, the Canny edge detection algorithm, and connected component labeling algorithms are all well-known techniques, and their specific methods will not be described here. Due to multiple layers of afterimages and contamination defects during the exposure process, the acquired image may appear blurry or distorted. Therefore, it is necessary to enhance the acquired image and obtain its edge information.

[0063] Step S002: Determine the degree of difference between each candidate region and the template based on the difference between each candidate region and the theoretical location region; based on the magnitude of the degree of difference between each candidate region and the template, select several candidate marking regions from all candidate regions.

[0064] It should be noted that: Further, it is necessary to remove interference information from the material markings and obtain the ghosting areas of the material markings. During the exposure process, dust or oil may appear on the glass substrate, forming fake small areas. By comparing these with the real marking template, the contamination points have low similarity due to significant differences in size and shape, and their edge morphology also differs considerably from the real markings. To eliminate these fake areas, since the current real markings are known, the enhanced image can be segmented based on the edge graphics. Parameters such as the area and aspect ratio of each suspected area can be extracted and compared with the preset real marking template to remove these fake areas and determine the areas where the markings exist.

[0065] It should be further explained that during the lithography process, due to multi-layer superposition or ghosting, multiple candidate markers may appear at the same location, meaning one candidate region contains another. When candidate markers overlap or intersect, ghosting regions will form in the image. Candidate regions can be judged by their spatial relationship; if intersections or partial overlaps are detected, it can be determined that these candidate regions collectively constitute a ghosting region.

[0066] Preferably, in one embodiment of the present invention, the method for obtaining the candidate marker region includes:

[0067] Obtain the area and aspect ratio of the theoretical location region of the real marker template in the marker image, denoted as the standard area and standard aspect ratio, respectively.

[0068] The real marker template is a rectangle, and the theoretical location area of ​​the real marker template in the marker image is also a matrix, with the aspect ratio being the ratio of the length of the long side to the length of the short side of the rectangle.

[0069] In the marked image, obtain the absolute value of the difference between the area of ​​each candidate region and the standard area. The normalized value is denoted as the first difference value. The absolute value of the difference between the aspect ratio of the smallest bounding rectangle of each candidate region and the standard aspect ratio is obtained. The normalized value is denoted as the second difference value. The mean of the first difference value and the second difference value is denoted as the region dissimilarity between each candidate region and the template.

[0070] It should be noted that in this embodiment, the following is used: and As respectively and The normalized value, where, This is a linear normalization function used to normalize data values ​​to a range of 0 to 1. Therefore, the closer the region dissimilarity between each candidate region and the template is to 0, the more similar it is to the true labeled template, and the greater the probability that it is a labeled tag.

[0071] It should be further noted that during actual imaging, some markers may appear blurred due to underexposure or overexposure. However, if the edge of a region belongs to a ghosted area of ​​the real marker, the displayed edge will show a high similarity to the corresponding edge of the real marker, indicating that the edge may be a ghosted part. To address the issue of missing edges in some candidate regions, the edge contour of any candidate region can be extracted, and its edge length calculated. An edge segment of the same length is extracted from the standard edge of the real marker template, and sliding matching is performed along the edge of the real template. The segment with the highest similarity is selected as the basis for determining the real edge of the candidate region.

[0072] Obtain the boundary length of each candidate region, and denote it as the target length.

[0073] Obtain the boundary length of the theoretical location region of the real marker template in the marker image, and denote it as the standard length.

[0074] When the target length is less than or equal to the standard length, on the boundary of each candidate region, starting from any pixel, traverse the entire sequence of pixels in a clockwise direction to form a pixel sequence, denoted as the target sequence. On the boundary of the theoretical location region of the real marker template in the marker image, starting from the i-th pixel, traverse the entire sequence of pixels in a clockwise direction to obtain a pixel sequence of length equal to the target length, denoted as the reference sequence of the i-th pixel on the boundary of the theoretical location region.

[0075] Using the DTW algorithm, the DTW distance of each pixel on the boundary between the target sequence and the theoretical location region is obtained. The minimum value of the DTW distance of all pixels on the boundary between the target sequence and the theoretical location region is obtained and denoted as the boundary dissimilarity between each candidate region and the template.

[0076] It should be noted that the DTW algorithm (Dynamic Time Warping) is a well-known technique, and its specific method will not be described here. In this embodiment, the distance between pixels is used as the matching distance between elements when matching two sequences. The smaller the DTW distance, the more similar the two sequences are. Therefore, the larger the minimum DTW distance among all pixels of the reference sequence on the boundary of the target sequence and the theoretical location region, the less similar each candidate region is to the boundary of the theoretical location region.

[0077] When the target length is greater than the standard length, starting from any pixel, the pixel sequence is formed by traversing the theoretical location region of the real marker template in the marker image in a clockwise direction, and this sequence is denoted as the new target sequence. On the boundary of each candidate region, starting from the j-th pixel, the pixel sequence is traversed in a clockwise direction to obtain a pixel sequence of the standard length, which is denoted as the new reference sequence for the j-th pixel on the boundary of each candidate region.

[0078] Using the DTW algorithm, the DTW distance between the new target sequence and the new reference sequence at each pixel on the boundary of each candidate region is obtained. The minimum value among the DTW distances between the new target sequence and the new reference sequence at all pixels on the boundary of each candidate region is recorded as the boundary dissimilarity between each candidate region and the template.

[0079] Obtain the boundary dissimilarity between each candidate region and the template. The normalized value and the mean of the regional dissimilarity between each candidate region and the template are denoted as the difference between each candidate region and the template.

[0080] Among them, with As The normalized value. The smaller the difference, the more likely the candidate region is to be a labeled region.

[0081] The preset difference threshold is 0.2, and this will be used as an example for explanation.

[0082] Among all candidate regions, those with differences less than the preset difference threshold are all marked as candidate regions.

[0083] Step S003: Based on the boundary symmetry of each candidate marking region and the overlap between candidate marking regions, determine the probability that each candidate marking region is a real marking region.

[0084] It should be noted that: since the ghosting region is formed by the overlap of multiple markers, if the candidate marker regions overlap, then the entire candidate marker region is determined to constitute the ghosting region. If there is no ghosting, then the region with the smallest overall difference among the candidate marker regions is determined as the current real marker.

[0085] It should be further noted that during the imaging process of the exposure machine, the edges of real material markers exhibit relatively stable reflection and absorption under illumination due to the stability of their material and surface properties, resulting in clear edge gradients and high contrast. Ghost images or afterimages, on the other hand, originate from the superposition of multiple layers of graphics or historical residues. Their overall intensity is usually lower, and the edges often exhibit occlusion or incompleteness. Since markers are mostly geometric structures such as matrices, circles, or crosses, they typically possess central symmetry characteristics. In this embodiment, rectangles are used, thus real markers can maintain a clear centrally symmetrical edge in the imaging. Ghost image edges formed by multi-layer superposition or afterimages, however, only retain some geometric features of the real marker, making it difficult to maintain overall symmetry. Therefore, the stronger the central symmetry of the edges retained in the candidate marker area, the greater the likelihood that it is a real marker.

[0086] Preferably, in one embodiment of the present invention, the method for obtaining the probability that each candidate marker region is a real marker includes:

[0087] Obtain the center-symmetric boundary of the boundary of each candidate marked region.

[0088] Specifically, for each candidate marking region, the boundary of the candidate marking region is rotated 180 degrees around the center point of the candidate marking region and used as the center-symmetric boundary. This is a well-known technique.

[0089] Obtain the number of overlapping pixels between the boundary of each candidate marker region and the central symmetric boundary. The normalized value is denoted as the target quantity.

[0090] The more overlapping pixels there are, the more centrally symmetric the current marker is, and the greater the likelihood that it is a true marker. As The normalized value.

[0091] Obtain the mean gradient value of all pixels on the boundary of each candidate marking region. The normalized value of is denoted as the first mean.

[0092] Among them, the larger the gradient value of the pixels on the boundary, the clearer the boundary. As The normalized value.

[0093] Obtain the absolute value of the difference between the boundary length of each candidate marker region and the boundary length of the theoretical location region. The inverse proportional normalized value is denoted as the boundary length consistency.

[0094] Among them, with As The inversely proportional normalized value. That is, the larger the third difference value, the more similar the lengths of the two boundaries.

[0095] The number of targets, the first mean, and the mean of the consistency of boundary length are denoted as the boundary clarity of each candidate marked region.

[0096] It should be noted that slight optical and mechanical deviations exist during exposure or imaging. The resulting ghosting, caused by these deviations, tends to be distributed around the true center. Regions with smaller overall offsets, and whose centers are closer to the expected center, have a more stable optical projection relationship with the true mark and are therefore more likely to correspond to the actual mark. Conversely, suspected mark regions with larger overall offsets and centers significantly deviating from the expected center are more likely to be ghostings caused by optical or mechanical errors.

[0097] Obtain the pixel point corresponding to the mean coordinates of the center points of all candidate marking regions, and record it as the estimated center point of the true mark.

[0098] It should be noted that in this embodiment, a coordinate system is constructed with the lower left corner of the marker image as the origin, the horizontal direction to the right as the positive direction of the horizontal axis, and the vertical direction upward as the positive direction of the vertical axis. The coordinates of each pixel in the marker image are obtained from the coordinate system.

[0099] Obtain the distance between the center point of each candidate marker region and the estimated center point of the actual marker, and record it as the center deviation value of each candidate marker region.

[0100] Get the overlapping area of ​​all candidate marked regions.

[0101] Specifically, the coordinates of all pixels within each candidate marker region are obtained, forming a coordinate set. The intersection of the coordinate sets of all candidate marker regions is obtained, and the pixels corresponding to all coordinates in the intersection form a region, which is used as the overlapping region.

[0102] Obtain the ratio of the area of ​​the overlapping region to the area of ​​each candidate marker region, denoted as the overlap ratio. Also obtain the center deviation value of each candidate marker region. The inversely proportional normalized value and the mean of the overlap ratio are denoted as the regional non-deviation degree of each candidate marked region.

[0103] Among them, with As The inversely proportional normalized value. That is, the greater the overlap ratio and the smaller the center deviation value, the more likely the candidate labeled region is a true label.

[0104] The mean of the boundary clarity and the region non-deviation of each candidate marking region is obtained and recorded as the probability that each candidate marking region is a true marking region.

[0105] Step S004: Obtain the real mark region based on the probability that each candidate mark region is a real mark; adjust the position of the exposure lamp based on the coordinate position of the real mark region in the mark image.

[0106] Among all the possibilities that the candidate marking region is the real marking region, the candidate marking region corresponding to the maximum value is recorded as the real marking region.

[0107] If there are multiple candidate marking regions corresponding to the maximum value, then all candidate marking regions corresponding to the maximum value are combined into the actual marking region.

[0108] The position of the exposure lamp is adjusted based on the coordinates of the actual marked area in the marked image, and then the exposure operation is performed.

[0109] Specifically: (1) Marking center deviation calculation: Calculate the deviation between the actual marking area and the preset reference position, including horizontal displacement, vertical displacement and rotational deviation. (2) Exposure lamp position adjustment: Based on the calculated deviation, control the exposure platform or exposure lamp to make fine adjustments so that the exposure lamp is accurately aligned with the material mark. (3) Exposure operation: When the exposure lamp position is aligned with the material mark, start the exposure process to evenly irradiate the material surface with light energy to achieve high-precision exposure. (4) Exposure completion and data recording: Record the exposure parameters, lamp position, material position and deviation information for subsequent quality tracking and system optimization.

[0110] This invention is now complete.

[0111] In summary, in this embodiment of the invention, the theoretical position region of the actual marker template in the marker image and several candidate regions in the marker image are obtained. Based on the difference between each candidate region and the theoretical position region, several alternative marker regions are selected from all candidate regions. Based on the boundary symmetry of each alternative marker region and the overlap between alternative marker regions, the actual marker region is obtained, which is used to adjust the position of the exposure lamp. This invention can accurately adjust the position of the exposure lamp to precisely expose the material.

[0112] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An exposure lamp intelligent control method for an exposure machine, characterized in that, The method comprises the following steps: Obtaining a theoretical position area of a real marker template in a marker sign image and a plurality of candidate areas detected by edge detection in the marker sign image; According to the difference between each candidate area and the theoretical position area, determining the difference degree of each candidate area and the template; according to the size of the difference degree of each candidate area and the template, screening a plurality of alternative marker areas from all candidate areas; According to the boundary symmetry of each alternative marker area and the overlapping condition between the alternative marker areas, determining the possibility of each alternative marker area being a real marker; According to the size of the possibility of each alternative marker area being a real marker, obtaining a real marker sign area; based on the coordinate position of the real marker sign area in the marker sign image, adjusting the position of the exposure lamp; The specific steps of determining the difference degree of each candidate area and the template include the following: According to the area and the length-width difference of each candidate area and the theoretical position area, determining the area dissimilarity of each candidate area and the template; Obtaining the boundary length of each candidate area, denoted as the target length; Obtaining the boundary length of the theoretical position area, denoted as the standard length; When the target length is less than or equal to the standard length, according to the matching similarity of the boundary of each candidate area on the boundary of the theoretical position area, determining the boundary dissimilarity of each candidate area and the template; When the target length is greater than the standard length, according to the matching similarity of the boundary of the theoretical position area on the boundary of each candidate area, determining the boundary dissimilarity of each candidate area and the template; Obtaining the normalized value of the boundary dissimilarity of each candidate area and the template and the mean value of the area dissimilarity, denoted as the difference degree of each candidate area and the template.

2. The intelligent exposure lamp control method for an exposure machine according to claim 1, wherein, The specific steps of determining the area dissimilarity of each candidate area and the template according to the area and the length-width difference of each candidate area and the theoretical position area include the following: Obtaining the area and the length-width ratio of the theoretical position area, denoted as the standard area and the standard length-width ratio respectively; Obtaining the normalized value of the absolute value difference of the area of each candidate area and the standard area, denoted as the first difference value, obtaining the normalized value of the absolute value difference of the length-width ratio of the minimum circumscribed rectangle of each candidate area and the standard length-width ratio, denoted as the second difference value, and taking the mean value of the first difference value and the second difference value as the area dissimilarity of each candidate area and the template.

3. The method of claim 1, wherein the method further comprises: The specific steps of determining the boundary dissimilarity of each candidate area and the template when the target length is less than or equal to the standard length include the following: On the boundary of each candidate area, starting from any pixel point, a pixel point sequence is formed by traversing one round in the clockwise direction, denoted as the target sequence; On the boundary of the theoretical position area, starting from the i-th pixel point, a pixel point sequence with a length of the target length is obtained by traversing one round in the clockwise direction, denoted as the reference sequence of the i-th pixel point on the boundary of the theoretical position area; The minimum value in the DTW distances of the reference sequence of all pixel points on the boundary of the target sequence and the theoretical position region is obtained, and is recorded as the boundary dissimilarity of each candidate region and the template.

4. The intelligent exposure lamp control method for an exposure machine of claim 1, wherein, When the target length is greater than the standard length, the boundary dissimilarity of each candidate region and the template is determined according to the matching similarity of the boundary of the theoretical position region on the boundary of each candidate region, and the specific steps include the following: On the boundary of the theoretical position region, a pixel point sequence is formed by starting from an arbitrary pixel point and traversing the pixel points in a clockwise direction, and is recorded as a new target sequence; On the boundary of each candidate region, a pixel point sequence with a standard length is obtained by starting from the jth pixel point and traversing the pixel points in a clockwise direction, and is recorded as a new reference sequence of the jth pixel point on the boundary of each candidate region; The minimum value in the DTW distances of the new target sequence and the new reference sequence of all pixel points on the boundary of each candidate region is obtained, and is recorded as the boundary dissimilarity of each candidate region and the template.

5. The intelligent exposure lamp control method for an exposure machine of claim 1, wherein, The specific steps of screening a plurality of candidate marking regions from all candidate regions include the following: In the difference degrees of all candidate regions and the template, all candidate regions less than a preset difference threshold are recorded as candidate marking regions.

6. The intelligent exposure lamp control method for an exposure machine of claim 1, wherein, The specific steps of determining the possibility of each candidate marking region being a real marker include the following: A center-symmetrical boundary of the boundary of each candidate marking region is obtained; A normalized value of the number of overlapping pixel points between the boundary of each candidate marking region and the center-symmetrical boundary is obtained, and is recorded as a target number; A normalized value of the mean value of the gradient values of all pixel points on the boundary of each candidate marking region is obtained, and is recorded as a first mean value; An inverse-proportion normalized value of the absolute value of the difference between the boundary length of each candidate marking region and the boundary length of the theoretical position region is obtained, and is recorded as a boundary length consistency; The boundary distinctness of each candidate marking region is determined according to the target number, the first mean value, and the boundary length consistency; The region non-deviation degree of each candidate marking region is determined according to the overlapping condition between the candidate marking regions; The mean value of the boundary distinctness and the region non-deviation degree of each candidate marking region is obtained, and is recorded as the possibility of each candidate marking region being a real marker.

7. The method of claim 6, wherein the method further comprises: The specific steps of determining the boundary distinctness of each candidate marking region according to the target number, the first mean value, and the boundary length consistency include the following: The mean value of the target number, the first mean value, and the boundary length consistency is recorded as the boundary distinctness of each candidate marking region.

8. The method of claim 6, wherein the method further comprises: The specific steps of determining the region non-deviation degree of each candidate marking region according to the overlapping condition between the candidate marking regions include the following: A pixel point corresponding to the mean coordinate of the coordinates of the center points of all candidate marking regions is obtained, and is recorded as an estimated center point of the real marker; The distance between the center point of each candidate marking region and the estimated center point of the real marker is obtained, and is recorded as a center deviation value of each candidate marking region; An overlapping region of all candidate marking regions is obtained. The ratio of the area of the overlapping region to the area of each candidate marking region is obtained, denoted as an overlapping proportion, and the inverse proportional normalized value of the center deviation value of each candidate marking region is obtained, denoted as a region non-deviation degree of each candidate marking region.

9. The intelligent control method for exposure lamps in an exposure machine according to claim 1, characterized in that, The specific steps of obtaining the real marking sign region include the following: Among the probabilities that all candidate marking regions are real marks, the candidate marking region corresponding to the maximum value is denoted as the real marking sign region.

Images