A monitoring operation and maintenance method and system of a photovoltaic power station

Abstract

The application discloses a kind of photovoltaic power station's monitoring operation and maintenance method and system, it is related to image processing technical field, including: to ground part registration image is corrected, obtains ground part perspective correction image;From ground part perspective correction image, the image area including ground mat, occlusion position and occlusion peripheral area is intercepted, to obtain ground part correction image;According to ground part correction image, obtain roll-over connection edge;The space distance between adjacent positions in each roll-over material strip is accumulated, to obtain the material extension length of each roll-over material strip;According to material extension length, obtain pre-rolling occupancy area;According to pre-rolling occupancy area, the region overlap extraction of current occlusion tooth area is carried out, to obtain across old mark occlusion receiving area;The application can reduce the influence of inclined shooting and local shielding on boundary identification.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a monitoring and maintenance method and system for photovoltaic power plants. Background Technology

[0002] In photovoltaic power plants, photovoltaic modules typically engage with the module frame via grounding pads. The interlocking teeth break through the insulation layer on the frame surface and maintain grounding conductivity. After long-term outdoor operation, maintenance disassembly, or loosening of the grounding pads, inspectors need to check the interlocking area, old interlocking tooth marks, and rolled-up aluminum sheets to determine whether the grounding pads remain in their original position or have been moved to a new position on the frame surface.

[0003] Current grounding pad inspections mainly rely on single-direction images and manual observation. Old bite marks, current bite indentations, and rolled-up aluminum sheets are easily obscured in two-dimensional images. Tilted shooting also changes the edge shape and distance relationship, making it difficult for inspectors to accurately determine the current bite position or reinstallation position of the grounding pad, thus creating a technical problem of inaccurate grounding pad maintenance position judgment. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies that make it difficult for inspection personnel to accurately determine the current engagement position or reinstallation position of the grounding pad, and to propose a monitoring and maintenance method and system for photovoltaic power plants.

[0005] To address the problems existing in the prior art, the present invention adopts the following technical solution:

[0006] A monitoring and operation and maintenance method for a photovoltaic power plant, comprising:

[0007] S1. Perform perspective correction on the registration image of the grounding part to obtain the perspective-corrected image of the grounding part; extract the image area including the grounding pad, the bite position and the bite periphery from the perspective-corrected image of the grounding part to obtain the grounding part correction image.

[0008] S2. Based on the grounding location correction image, obtain the rolled-up connection edge;

[0009] S3. Accumulate the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, obtain the pre-rolling occupied area;

[0010] S4. Based on the area occupied before the rollover, the current occlusal tooth area is extracted by regional overlap to obtain the cross-old trace occlusal bearing area.

[0011] S5. Extract the area closest to the current biting tooth area to obtain the target bearing area; based on the cross-old mark biting bearing area and the target bearing area, monitor and maintain the grounding pad of the target photovoltaic module.

[0012] To address the aforementioned problems, the present invention also provides a monitoring and operation and maintenance system for a photovoltaic power plant, the system comprising:

[0013] The calibration module performs perspective correction on the registration image of the grounding part to obtain a perspective-corrected image of the grounding part; it then extracts an image region including the grounding pad, the occlusion position, and the occlusion periphery from the perspective-corrected image of the grounding part to obtain a grounding part calibration image.

[0014] The image correction module, based on the grounding location, obtains the image of the image being flipped and connected.

[0015] The in-situ restoration module accumulates the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, the pre-rolling occupied area is obtained;

[0016] The occlusal reconstruction module extracts the overlapping areas of the current occlusal tooth area based on the pre-rollover occupied area, and obtains the cross-old trace occlusal bearing area.

[0017] The operation and maintenance positioning module extracts the area closest to the current biting tooth area to obtain the target bearing area; based on the biting bearing area across old marks and the target bearing area, it monitors and maintains the grounding pads of the target photovoltaic module.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. By acquiring images from multiple directions, registering images, and correcting perspective, and combining spatial depth information of the corresponding locations, a grounding correction image is formed. This ensures that the bite indentation, old bite marks, and rolled aluminum sheet maintain their corresponding positions under a unified coordinate system, reducing the impact of tilted shooting and partial obstruction on boundary recognition, and providing verifiable image evidence for judging the actual bite state of the grounding pad.

[0020] 2. By extracting the old tooth marks, rolled aluminum sheets, and rolled connecting edges, the rolled material is divided into strips and the material extension length is accumulated according to spatial coordinates. Then, each strip is restored in reverse from the starting point of the spread. This forms the pre-rolled area and the old tooth mark backfill area, so that the rolled aluminum material is restored to a comparable original area, and the material connection relationship when the current biting tooth crosses the old mark is clarified.

[0021] 3. By removing old bite marks and pre-rollover occupied areas from the component frame, and filtering areas that can completely accommodate the bite teeth based on the outer contour of the current bite tooth area, and then selecting the target receiving area with the smallest distance from the current bite tooth area, the current bite position and reinstallation position can be determined, forming a position adjustment relationship and reducing misjudgment of the grounding pad maintenance position. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0023] Figure 1 This is a flowchart illustrating a monitoring and maintenance method for a photovoltaic power station according to an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0025] This embodiment provides a monitoring and maintenance method for a photovoltaic power station. (See also...) Figure 1 Specifically, including:

[0026] In an embodiment of the present invention, obtaining a perspective-corrected image of the grounding portion includes:

[0027] S1. Perform perspective correction on the registration image of the grounding part to obtain the perspective-corrected image of the grounding part; extract the image area including the grounding pad, the bite position and the bite periphery from the perspective-corrected image of the grounding part to obtain the grounding part correction image.

[0028] The perspective correction image of the grounding part uses the surface of the module frame where the grounding pad is located in the target photovoltaic module as the correction reference. After the image positions formed by different shooting directions are matched, the image formed by the tilt shooting is eliminated to eliminate the side length shrinkage and angle offset. The perspective correction image of the grounding part retains the pixel content of the grounding pad, the contact part between the grounding pad and the module frame, and the surface of the adjacent module frame. It is used to provide a unified image coordinate reference for the spatial depth recovery of the grounding part and the positioning of the interlocking area.

[0029] Multi-directional image acquisition is performed on the location of the grounding pad in the target photovoltaic module to obtain multi-view images of the grounding part;

[0030] The target photovoltaic module is a photovoltaic module designated for grounding connection status monitoring and equipped with grounding pads during photovoltaic power station inspection tasks. The grounding pad is a conductive connection component placed between the module frame and the grounding conductive connector of the target photovoltaic module and pressed into the surface of the module frame through interlocking teeth. It is used to break the insulation layer on the surface of the module frame and maintain grounding conductivity. Multi-directional image acquisition is the process of acquiring images from different imaging directions around the grounding pad to cover the same grounding part. The multi-view image of the grounding part is a series of images organized according to the acquisition sequence of each imaging direction. Each image contains the same grounding pad and the module frame area where the grounding pad is located, and carries the imaging direction and image size information at the time of acquisition.

[0031] Specifically, the photovoltaic modules with grounding pads to be monitored, as specified in the photovoltaic power station inspection task, are identified as the target photovoltaic modules. The geometric center of the grounding pad in the target photovoltaic module is used as the center of the imaging area. The image acquisition device is set up in a position that can simultaneously cover the grounding pad and the surface of the module frame around the grounding pad. The imaging direction of the image acquisition device is changed sequentially along the circumference of the grounding pad while keeping the center of the imaging area unchanged in each imaging direction. An image of the grounding part, including the complete grounding pad and the surface of the module frame around the grounding pad, is acquired in each imaging direction. The imaging direction, image width, image height, and acquisition order recorded by the image acquisition device are written into the corresponding grounding part acquisition image. When the grounding part acquisition image does not completely cover the grounding pad or the surface of the module frame, the distance between the image acquisition device and the grounding pad is adjusted along the original imaging direction and the grounding part acquisition image is reacquired. The grounding part acquisition images corresponding to each imaging direction are combined into a multi-view image of the grounding part according to the acquisition order.

[0032] Extract the positions of the grounding pad edge and component frame from multi-view images of the grounding part under different shooting directions;

[0033] The edge of the grounding pad under different shooting directions is the closed boundary position between the grounding pad pixel and the adjacent non-grounding pad pixel in each image of the multi-view image of the grounding part, and is represented according to the pixel coordinates of the corresponding image; the component frame position is the set of pixels belonging to the surface of the target photovoltaic component frame and its boundary coordinates in each image of the multi-view image of the grounding part, and the component frame position surrounds or carries the area where the grounding pad is located under the corresponding shooting direction.

[0034] Specifically, the multi-view images of the grounding area are processed frame by frame according to the acquisition sequence. The red, green, and blue channel values ​​of each pixel in each grounding area image are added together and divided by 3 to form the pixel grayscale value of each grounding area image. For each non-image boundary pixel, the absolute values ​​are subtracted from the grayscale values ​​of the left and right adjacent pixels, and subtracted from the grayscale values ​​of the upper and lower adjacent pixels. These two absolute values ​​are then added together to form the pixel edge change. For image boundary pixels, the pixel edge change is taken from the non-image boundary pixel closest to the image boundary pixel. All pixel edge changes in the same grounding area image are added together and divided by the total number of pixels to form the average image edge change. Pixel edge changes greater than or equal to the average image edge change are considered average edge changes. Pixels with varying edge values ​​are connected according to their adjacency to form candidate edge contours. The candidate edge contours are arranged from most to least number of pixels. The region enclosed by the candidate edge contour with the most pixels that extends to the boundary of the grounding part image is determined as the component border position. The candidate edge contour located inside the component border position, forming a closed region, and having the smallest distance from the center of the imaging area of ​​the grounding part image is determined as the grounding pad edge. When the candidate edge contour cannot simultaneously form the grounding pad edge and the component border position, the corresponding grounding part image is returned to the aforementioned acquisition process, and the exposure is adjusted step by step along the same imaging direction until the grounding pad edge and the component border position are formed. The grounding pad edge and component border positions corresponding to each grounding part image are collected according to the shooting direction to obtain the grounding pad edge and component border positions under different shooting directions.

[0035] The pixel grayscale value unifies the three color channels of the same pixel into directly comparable values. The pixel edge change reflects the brightness abrupt changes in the left-right and up-down directions of the pixel. The average edge change of the image is formed by all pixels of the current image and is used to distinguish the boundary between continuous surfaces and objects. Combined with the spatial relationship that the grounding pad is located in the center of the imaging area and the component frame covers the outer area during acquisition, the position of the grounding pad edge and the component frame can be determined from the same image coordinate reference in each shooting direction.

[0036] Based on the positions of the grounding pad edge and component frame under different shooting directions, image registration is performed on multi-view images of the grounding part to obtain the grounding part registration image;

[0037] Image registration is the process of transforming pixels representing the same grounding pad and the same component frame position under different shooting directions to a unified reference image coordinate. The grounding part registration image is a multi-channel image formed by using the reference image coordinates in the multi-view images of the grounding part as the common coordinates and aligning the grounding pad edge and component frame position in the other shooting directions with the same position in the reference image. The grounding part registration image retains the pixel source correspondence of each shooting direction.

[0038] Specifically, the grounding location image acquired first in the multi-view images of the grounding location is designated as the registration reference image. From the registration reference image and the component border positions of each of the remaining grounding location images, the component border inflection points where the contour direction changes are extracted. Furthermore, from the edge of the grounding pad, the grounding pad contour positions with the smallest, largest, smallest, and largest abscissas are extracted. Based on the order of the component border inflection points along the component border positions and the directional names of the grounding pad contour positions, the component border inflection points in each of the remaining grounding location images are then... The outline position of the grounding pad is matched one by one with the component border turning points and the outline position of the grounding pad in the registration reference image to form registration corresponding points. If the image of any grounding part cannot form 4 sets of registration corresponding points that are not on the same straight line, the image of the corresponding grounding part is returned to the previous acquisition process and reacquired. For each set of registration corresponding points, the abscissa of the source image, the ordinate of the source image and the abscissa of the reference image, and the ordinate of the reference image are respectively established as abscissa projection equations and ordinate projection equations. All abscissa projection equations and ordinate projection equations are arranged into a set of projection equations according to the order of the registration corresponding points. The image projection transformation coefficients are formed by sequentially selecting the unprocessed coefficient with the largest absolute value as the elimination reference, subtracting the reference equation multiplied by the corresponding coefficient ratio from each non-reference equation, and substituting the result in reverse. When the elimination reference is 0, the equations containing non-zero corresponding coefficients are swapped to continue the elimination process. When no non-zero corresponding coefficients exist, the image acquired at the corresponding grounding location is returned to the aforementioned registration corresponding point formation process. For each pixel of each of the remaining grounding location acquired images, the source image's abscissa and ordinate are sequentially substituted into the abscissa projection equation and ordinate projection equation defined by the image projection transformation coefficients to form each... The registration coordinates of the source image pixels in the registration reference image are determined. When the registration coordinates are not integer coordinates, the product weights of the four integer coordinate pixels are determined according to the horizontal and vertical distances between the registration coordinates and the four surrounding integer coordinates. The channel values ​​of the four integer coordinate pixels are multiplied by their corresponding product weights and then added together to form the pixel channel values ​​of the registration coordinates. The positions where the registration coordinates exceed the range of the registration reference image are marked as regions without corresponding pixels. The registration reference image, the pixel channel values ​​of the registration coordinates corresponding to the other shooting directions, and the regions without corresponding pixels are combined according to the shooting direction to form the grounding part registration image.

[0039] The component frame inflection point and the grounding pad outline position are respectively derived from the grounding pad edge and component frame position under different shooting directions and represent the same entity position. The four sets of registration corresponding points that are not on the same straight line can define the planar projection relationship. The elimination and reverse substitution of the projection equation set form the image projection transformation coefficient from each source image to the registration reference image. The pixel channel values ​​can be filled into the unified coordinate according to the distance weight combination of the four integer coordinates around the registration coordinates. Therefore, the same grounding parts in each shooting direction can be merged to form the grounding part registration image.

[0040] Extract the component border surface orientation from the registered image at the grounding location;

[0041] The component frame surface direction refers to two non-parallel plane directions that extend along the component frame surface in the registered image of the grounding part. The two plane directions are defined by two sets of opposite sides of the component frame position and are represented by the direction vector in the coordinate of the registration reference image. They are used to determine the perspective contraction direction of the component frame surface caused by tilted shooting.

[0042] Specifically, using the top-left pixel of the registration reference image in the registration image of the grounding part as the origin, the pixel row direction as the positive direction of the horizontal coordinate, and the pixel column direction as the positive direction of the vertical coordinate, each continuous straight line edge is extracted sequentially along the contour of the component border position. The horizontal component of the straight line edge is formed by subtracting the horizontal coordinate of the starting point from the horizontal coordinate of the end point of each continuous straight line edge, and the vertical component of the straight line edge is formed by subtracting the vertical coordinate of the starting point from the vertical coordinate of the end point of each continuous straight line edge. The square root of the sum of the squares of the horizontal and vertical components of the straight line edge is taken to form the length of the continuous straight line edge. The horizontal and vertical components of the continuous straight line edge with a length other than 0 are divided by the length of the continuous straight line edge to form the direction of the continuous straight line edge. The horizontal components of any two continuous straight line edge directions are cross-multiplied and subtracted. Continuous straight lines with an absolute value of 0 after the subtraction result are grouped into the same parallel edge group. The lengths of each continuous straight line edge in the edge group are added together to form the cumulative length of the parallel edge group. The parallel edge groups are arranged from largest to smallest cumulative length. The parallel edge group with the largest cumulative length is selected as the first surface edge group. Then, the parallel edge group with the largest cumulative length after cross-multiplication with the first surface edge group is selected as the second surface edge group. If no second surface edge group exists, the process of extracting the component border position is returned to re-form the component border position. The directions of each continuous straight line edge in the first and second surface edge groups are weighted and added together according to the length of the continuous straight line edge, and then divided by the corresponding cumulative length of the parallel edge group to form the first component border surface direction and the second component border surface direction. The first component border surface direction and the second component border surface direction are combined according to the order of the cumulative length of the parallel edge groups to form the component border surface direction.

[0043] The direction of the continuous straight edge is formed by the coordinate difference of the endpoints of the component frame position. The length of the continuous straight edge is used to eliminate the influence of different edge lengths on the direction combination. The result of the cross multiplication and subtraction is 0, which means that the two continuous straight edges are parallel. The two non-parallel parallel edge groups with the largest cumulative length correspond to the two main opposite edges on the component frame surface. The two main opposite edges can be combined according to the length of the continuous straight edge to form two extension directions on the component frame surface and determine the direction of the component frame surface.

[0044] Based on the surface orientation of the component frame, the registered image of the grounding part is subjected to perspective correction to obtain the perspective-corrected image of the grounding part.

[0045] Perspective correction is the process of mapping the non-parallel or shortened component border lines in the registered image of the ground part to mutually perpendicular correction coordinate directions based on the surface direction of the component border, and rearranging the pixels of the ground part according to the mapping relationship.

[0046] Specifically, from the component border positions of the grounding part registration image, two sides extending along the surface direction of the first component border and two sides extending along the surface direction of the second component border are selected respectively. The four component border corner points formed by the intersection of the two sets of sides are arranged in the order of upper left, upper right, lower right, and lower left. The pixel coordinate differences between the two sets of adjacent component border corner points along the surface direction of the first component border are squared, added together, and the square root is taken. The larger value of the two calculation results is selected as the width of the correction image. The pixel coordinate differences between the two sets of adjacent component border corner points along the surface direction of the second component border are squared, added together, and the square root is taken. The larger value of the two calculation results is selected as the height of the correction image. A rectangular correction coordinate system defined by the width and height of the correction image is established with the upper left corner of the correction image as the origin. The four sets of sides are then... The corner points of the component frame and the four corner points of the rectangular correction coordinates are arranged in directional order to form perspective correction corresponding points. The perspective correction projection coefficients are formed according to the composition of the aforementioned projection equation set, the selection of elimination reference, the equation elimination and reverse substitution process. The coordinates of each effective pixel in the registered image of the grounding part are substituted into the perspective correction projection coefficients to form perspective correction coordinates. For non-integer perspective correction coordinates, the horizontal and vertical distances between the four surrounding integer coordinates and the perspective correction coordinates are formed into product weights. The pixel channel values ​​of the four surrounding integer coordinates are multiplied by their corresponding product weights and then added together to form the pixel channel values ​​of the perspective correction coordinates. The pixel channel values ​​of all perspective correction coordinates are arranged according to the rectangular correction coordinates, and the pixel source correspondence of the grounding pad edge, component frame position and each shooting direction is preserved to obtain the perspective correction image of the grounding part.

[0047] The component border surface direction defines two actual extension directions of the component border surface. The four corner points of the component border and the four corner points of the rectangular correction coordinate define the projection relationship from the tilted plane to the front view rectangle. The width and height of the correction image are taken as the maximum value of the pixel distance between the corresponding two sides to retain the existing side length information. The perspective correction projection coefficient and distance weight continuously map the effective pixels of the ground part registered image to the rectangular correction coordinate, thus eliminating the directional deviation caused by tilt shooting and obtaining the perspective correction image of the ground part.

[0048] In an embodiment of the present invention, obtaining a grounding correction image includes:

[0049] The grounding correction image is formed by retaining the grounding pad, the bite position and the surrounding area of ​​the bite from the grounding perspective correction image, and associating spatial depth information with each corresponding image position of the retained area. The grounding correction image organizes color data and spatial depth information with perspective-corrected two-dimensional pixel coordinates, and is used to identify the depressions and rolled-up materials formed by the bite of the grounding pad.

[0050] Based on the corresponding positions between the multi-view images of the grounding part, obtain the spatial depth information of each corresponding position in the perspective correction image of the grounding part;

[0051] The corresponding positions between the multi-view images of the grounding part are the pixel positions representing the same entity point of the grounding part in images taken from different shooting directions, and are determined by the pixel source correspondence preserved in the grounding part registration image; the spatial depth information is the spatial distance from the entity point represented by the corresponding position in the perspective correction image of the grounding part to the acquisition position of the registration reference image. The spatial depth information is recorded using the calibrated length unit of the image acquisition device and associated with the perspective-corrected pixel coordinates.

[0052] Specifically, from the pixel source correspondence of each shooting direction preserved in the registered image of the grounding part, the corresponding positions representing the same entity point between the multi-view images of the grounding part are obtained one by one. The imaging center coordinates, imaging focal length, and pixel size corresponding to each shooting direction are obtained from the calibration data of the image acquisition device. The corresponding imaging direction vector is obtained from the imaging direction carried by the images acquired from each grounding part. The horizontal line-of-sight component is formed by subtracting the horizontal coordinate of the pixel at each corresponding position from the horizontal coordinate of the image center, multiplying by the pixel size, and dividing by the imaging focal length. The vertical line-of-sight component is formed by subtracting the vertical coordinate of the pixel at each corresponding position from the vertical coordinate of the image center, and multiplying by the image center focal length. The longitudinal line-of-sight component is formed by dividing the pixel size by the imaging focal length. The lateral line-of-sight component, the longitudinal line-of-sight component, and the imaging direction vector are added according to the calibration coordinates of the image acquisition device and divided by the square root of the sum of the squares of each component to form the unit line-of-sight direction for each corresponding position. Two unit line-of-sight directions are selected sequentially from different shooting directions. The two unit line-of-sight directions are multiplied by the unknown line-of-sight distance and added to the corresponding imaging center coordinates to form two spatial lines of sight. The coordinate components of the two spatial lines of sight are subtracted and then multiplied by the two unit line-of-sight directions respectively and added to form the two line-of-sight distances that minimize the square of the interval between the two spatial lines of sight. The equation is used to eliminate one unknown line-of-sight distance through cross-multiplication and subtraction. The result of the subtraction is then divided by the coefficient of the remaining unknown line-of-sight distance to form two line-of-sight distances. When the coefficient of the remaining unknown line-of-sight distance is 0, the current two unit line-of-sight directions are discarded, and unit line-of-sight directions from other shooting directions are selected. The two line-of-sight distances are substituted into the two spatial lines of sight to form two spatial positions. The two spatial positions are added according to their coordinate components, and then divided by 2 to form the spatial position of the corresponding entity point. The spatial position of the corresponding entity point is subtracted from the coordinates of the imaging center of the registration reference image component by component, and the squares of the differences between the coordinate components are summed and taken as the result. The square root is used to form the spatial depth of the corresponding position. The spatial depths of all corresponding positions formed by different shooting directions are added together and divided by the number of effective shooting direction combinations to form the average spatial depth of the corresponding position. When the number of effective shooting direction combinations is 0, the corresponding position is marked as a depth missing position and the next effective corresponding position is processed. The average spatial depth of the corresponding position is written into the corresponding position in the perspective correction image of the ground part according to the pixel source correspondence of the ground part registration image. The depth missing position is kept as a null value, thus obtaining the spatial depth information of each corresponding position in the perspective correction image of the ground part.

[0053] The imaging center coordinates, imaging direction, imaging focal length, and pixel size of each shooting direction convert the corresponding position in the image into a unit line of sight from different imaging centers. The same entity point should be located at the intersection of spatial lines of sight from different shooting directions. By minimizing the square of the interval between two spatial lines of sight and taking the midpoint of the coordinates of the two spatial positions, the effect of incomplete line of sight intersection caused by discrete pixels can be reduced. The distance from the spatial position of the corresponding entity point to the imaging center coordinates of the registration reference image is the spatial depth of the corresponding position in the perspective correction image of the ground part. The averaging of multiple sets of valid results can form the spatial depth information of each corresponding position in the perspective correction image of the ground part.

[0054] Determine the engagement position of the grounding pad;

[0055] The engagement position is the image position where the engagement teeth of the grounding pad press into the frame surface of the target photovoltaic module and form a local concave contact. The engagement position is located inside or connected to the edge of the grounding pad and is represented by the set of pixel coordinates in the perspective-corrected image of the grounding part.

[0056] Specifically, in the perspective correction image of the grounding part, pixels whose pixel area enclosed by the edge of the grounding pad coincides with the component frame are identified as potential biting positions. Spatial depth information associated with each potential biting position and the pixels adjacent to the component frame around it is obtained. When the spatial depth information of a potential biting position is null, neighboring pixels with spatial depth information are searched from the nearest to the farthest pixel in order of pixel distance, and the spatial depth information associated with the neighboring pixel with the smallest pixel distance is used to fill the potential biting position. When multiple neighboring pixels with the smallest pixel distance exist, the spatial depth information associated with these multiple neighboring pixels is added together, divided by the number of neighboring pixels, and then added to the potential biting position. The spatial depth information of each bite candidate position is compared with the spatial depth information of the surrounding component border positions. The bite candidate position with spatial depth information greater than the spatial depth information of all adjacent component border positions is determined as the bite recess position. Adjacent bite recess positions are connected to form a bite recess region. The pixel with the maximum spatial depth information in each bite recess region is taken as the bite center position. The bite recess region and the bite center position are jointly determined as the bite position corresponding to the grounding pad. When there is no bite recess position, the pixel with the maximum spatial depth information in all bite candidate positions and the pixel adjacent to the pixel with the maximum value and located within the edge of the grounding pad are determined as the bite position corresponding to the grounding pad.

[0057] The bite candidate position is defined by the physical contact relationship between the edge of the grounding pad and the position of the component frame. After the bite teeth of the grounding pad are pressed into the surface of the component frame, the corresponding physical point is moved away from the imaging center of the registration reference image relative to the surrounding component frame surfaces. Therefore, when the spatial depth information of the bite candidate position is greater than the spatial depth information of the adjacent component frame position, it indicates a local concavity. The bite position corresponding to the grounding pad can be determined by combining the bite concavity area and the pixels with the maximum spatial depth information in the bite concavity area.

[0058] Based on the biting position, extract the area surrounding the biting position from the component border to obtain the biting perimeter area;

[0059] The engagement perimeter area is a continuous component frame surface area within the component frame location that surrounds each engagement position and extends to the edge of the grounding pad, defining the contact range. The engagement perimeter area includes the non-recessed component frame surface adjacent to the engagement position and the area where material may roll over after the engagement teeth act.

[0060] Specifically, the component border position in the perspective correction image of the grounding part is taken as the area to be extracted, and the closed range enclosed by the edge of the grounding pad is taken as the bite contact boundary. Starting from the boundary pixel of each bite recessed area in the bite position, the adjacent pixels in the area to be extracted are checked layer by layer according to the upper side, lower side, left side, right side and four diagonally adjacent positions. The adjacent pixels located within the component border position and not crossing the bite contact boundary are included in the bite extension area. The next layer of adjacent pixels is checked with the newly included adjacent pixels in the bite extension area. When the next layer of adjacent pixels reaches the edge of the grounding pad or the boundary of the component border position, the check in the corresponding direction is stopped. The bite extension areas formed by all bite recessed areas are merged. The pixels connected to at least one bite recessed area are retained and the pixels where the grounding pad body is located are removed. The retained pixels are arranged according to the coordinates of the perspective correction image of the grounding part to obtain the bite perimeter area.

[0061] The grounding correction image is obtained by cropping an image region including the grounding pad, the bite position, and the bite periphery from the perspective correction image of the grounding part, and writing the spatial depth information into the corresponding position in the image region.

[0062] The image region is a continuous rectangular pixel range in the perspective correction image of the grounding part that can completely surround the grounding pad, the interlocking position, and the area surrounding the interlocking. The image region follows the pixel arrangement direction of the perspective correction image of the grounding part. Writing spatial depth information to the corresponding position means adding spatial depth data associated with the same entity position to each pixel in the image region, so that the color data and spatial depth data share the same pixel coordinates.

[0063] Specifically, the pixel coordinates of all pixels belonging to the grounding pad, the occlusion position, and the surrounding area in the perspective correction image of the grounding part are used to form candidate coordinates for the correction image. The horizontal and vertical coordinates of these candidate coordinates are compared, and the minimum, maximum, minimum, and maximum values ​​of the horizontal and vertical coordinates are selected to define a rectangular cropping boundary. Color channel data within this rectangular cropping boundary is extracted from the perspective correction image of the grounding part to form a cropped image of the grounding part. The spatial depth information of each corresponding position in the perspective correction image of the grounding part is found according to the original pixel coordinates in the cropped image. The average spatial depth of the found corresponding positions is written into the depth channel of the cropped image of the grounding part. When the depth corresponding to a position is missing from the original pixel coordinates, the depth information is obtained from the original pixel coordinates along the horizontal and vertical directions. To find the effective location with the smallest pixel distance and corresponding average spatial depth, the effective location distance is obtained by adding the squares of the differences between the effective location and the original pixel coordinates (x-coordinate and y-coordinate) and taking the square root. The average spatial depth of the effective location is then divided by the effective location distance to obtain the inverse-weighted depth. The inverse-weighted depths of all effective locations are summed and divided by the sum of the inverses of all effective location distances to obtain the missing location completion depth. When the sum of the inverses of all effective location distances is 0, the depth channel of the original pixel coordinates is left empty. The missing location completion depth is written into the corresponding depth channel of the grounding part cropped image. The color channel data, depth channel, grounding pad, bite position, and coordinate markers of the bite surrounding area of ​​the grounding part cropped image are combined to form the grounding part correction image.

[0064] The minimum, maximum, minimum, and maximum values ​​of the horizontal and vertical coordinates of the candidate coordinates of the corrected image define the smallest rectangular range that can completely encompass the grounding pad, the bite position, and the surrounding area. The original pixel coordinates ensure that the color channel data and spatial depth information maintain the same entity position correspondence. The depth missing position is calculated by combining the average spatial depth of the surrounding corresponding positions according to the inverse of the effective position distance, which avoids using distant pixels to replace nearby pixels. The image composed of the color channel data and the depth channel according to the same coordinates is the grounding correction image.

[0065] S2. Based on the grounding location correction image, obtain the rolled-up connection edge;

[0066] In an embodiment of the present invention, obtaining the rolled-up connecting edge includes:

[0067] The rolled-up connecting edge is the continuous boundary between the rolled-up aluminum sheet and the unrolled component frame surface, which is formed by the compression of the old bite mark by the bite teeth of the grounding pad. The rolled-up connecting edge is located at the root of the old bite mark rolled-up aluminum sheet and is represented by ordered pixel coordinates in the grounding part correction image. It is used to define the fixed side of the rolled-up material strip and the starting side of the reverse spreading of the material.

[0068] Extract the concave contour formed by the pressing of the occlusal teeth of the grounding pad from the peri-occlusal area;

[0069] The interlocking teeth of the grounding pad are sharp tooth structures that are set facing the component frame surface and pressed into the component frame surface under the fastening action; the recessed profile is a closed boundary in which the spatial depth of the interlocking perimeter area changes abruptly relative to the adjacent non-recessed component frame surface and forms a local recessed range. The recessed profile corresponds to the edge formed by the interlocking teeth of the grounding pad being pressed into the component frame surface.

[0070] Specifically, spatial depth information of pixels in the bite perimeter region is obtained one by one from the grounding correction image according to the coordinates of the bite perimeter region. For each pixel in the bite perimeter region, the spatial depth information of the left neighboring pixel is subtracted from the spatial depth information of the right neighboring pixel, and the absolute value is taken. Then, the spatial depth information of the upper neighboring pixel is subtracted from the spatial depth information of the lower neighboring pixel, and the absolute value is taken. The two absolute values ​​are then added together to form the perimeter depth change. When the spatial depth information of any neighboring pixel is null, the bite perimeter region pixel with spatial depth information closest to the corresponding neighboring pixel is used for subtraction. The total perimeter depth change is summed and divided by the number of bite perimeter region pixels with perimeter depth change to form the average perimeter depth change. When the number of pixels in the surrounding area of ​​the bite is 0, the spatial depth information writing process is returned. Pixels in the surrounding area of ​​the bite with a depth change greater than or equal to the average depth change are identified as candidate pixels for the concave boundary. Adjacent candidate pixels for the concave boundary are connected according to the arrangement relationship in the surrounding area of ​​the bite to form a candidate closed contour. The spatial depth information of the pixels inside each candidate closed contour is added and divided by the number of pixels inside to form the average depth inside the contour. The spatial depth information of the pixels at the edge of the adjacent component outside the candidate closed contour is added and divided by the number of adjacent pixels outside to form the average depth outside the contour. Candidate closed contours with an average depth inside the contour greater than the average depth outside the contour are retained in pixel coordinate order to form a concave contour formed by the bite teeth of the grounding pad.

[0071] The peripheral depth change reflects the abrupt change in surface height between adjacent pixels in the bite perimeter area. The peripheral average depth change is formed by the data of the current bite perimeter area itself and is used to exclude depth changes in the continuous plane. The average depth inside the candidate closed contour is greater than the average depth outside the candidate closed contour, indicating that the position surrounded by the candidate closed contour is recessed inward relative to the component border surface. Therefore, the retained candidate closed contour is the recessed contour formed by the bite teeth of the grounding pad.

[0072] Based on the concave contour, the old bite marks are obtained;

[0073] Old bite marks are historical bite areas formed by the recessed contour and the recessed area of ​​the component frame surrounded by the recessed contour. Old bite marks correspond to bite indentations formed on the grounding pad before the current installation position and still remain on the surface of the component frame.

[0074] Specifically, the image is processed row by row according to the vertical coordinate of the concave contour in the grounding part correction image from small to large. For each pixel row, all the horizontal coordinates of the concave contour and the pixel row are obtained. All the horizontal coordinates are arranged from small to large and grouped together in the order of arrangement. The pixels between two horizontal coordinates in each group and located within the component border position are marked as pixels inside the contour. When the number of intersecting horizontal coordinates in the same pixel row is odd, the horizontal coordinates that are missing pairing at the end of the arrangement are discarded and the next pixel row is processed. The pixels inside the contour formed by each pixel row are merged with the corresponding concave contour to form a concave filling area. Different concave filling areas are connected according to the adjacent relationship. The concave filling area that does not coincide with the bite position corresponding to the grounding pad and is located in the bite periphery area is determined as the historical concave area. The historical concave area and the concave contour corresponding to the historical concave area together form the old bite mark.

[0075] When a closed concave contour intersects with the same pixel row, the pixels between two adjacent intersection points are located inside the concave contour after being arranged by the horizontal coordinate. Filling row by row can convert the boundary coordinates into a complete concave area. The surrounding area of ​​the bite limits the search range of historical indentations. Removing the concave filling area that coincides with the bite position corresponding to the grounding pad can distinguish between the current bite area and the historical bite area. Therefore, the historical concave area and the corresponding concave contour together form the old bite marks.

[0076] The old bite marks were extracted in sheet-like areas to obtain the old bite mark rolled aluminum sheet;

[0077] Sheet-shaped region extraction refers to the separation of continuous aluminum material regions from the edge of old bite marks, where the thickness variation in the direction of the region is less than that in the direction of surface extension, and which are raised outward relative to the surface of the unrolled component frame. Old bite mark rolled aluminum sheet refers to sheet-shaped aluminum material that was squeezed out of the component frame surface by the bite teeth when the old bite marks were formed, and is still connected to the component frame position at the root. The old bite mark rolled aluminum sheet is located outside the edge of the old bite mark.

[0078] Specifically, starting from the outer boundary pixel of the old bite mark, layer by layer, patch-shaped candidate pixels adjacent to the old bite mark but not belonging to it are obtained in the bite perimeter region. From each patch-shaped candidate pixel, the component border position pixel closest to the patch-shaped candidate pixel but not belonging to it is obtained in the direction away from the old bite mark. The spatial depth information of the component border position pixel is used as the border reference depth of the corresponding patch-shaped candidate pixel. The spatial depth information of the patch-shaped candidate pixel is subtracted from the border reference depth to form the patch-shaped protrusion depth. Patch-shaped candidate pixels with a patch-shaped protrusion depth greater than 0 are connected according to their adjacency relationship to form a patch-shaped protrusion region. The values ​​of each patch-shaped candidate pixel are then statistically analyzed. The number of pixels continuously distributed along the boundary of the old bite mark and the number of pixels continuously distributed outward from the boundary of the old bite mark are used to determine the sheet-like protrusions that are connected to the boundary of the old bite mark. The sheet-like protrusions that are continuously distributed along the boundary of the old bite mark and have a greater number of pixels continuously distributed outward from the boundary of the old bite mark are defined as the rolled aluminum sheet areas. When the spatial depth information or the reference depth of the frame of the sheet-like candidate pixel is null, the frame of the component with spatial depth information that is closest to the sheet-like candidate pixel is used for comparison in the same direction. All rolled aluminum sheet areas are arranged according to the coordinates in the grounding correction image to obtain the old bite mark rolled aluminum sheet.

[0079] The border reference depth comes from the position of the component border outside the sheet-like candidate pixel that has not been rolled up. The sheet-like protrusion depth formed by subtracting the spatial depth information of the sheet-like candidate pixel from the border reference depth is greater than 0, which means that the sheet-like candidate pixel is raised relative to the surface of the component border towards the imaging center. The number of pixels continuously distributed along the boundary of the old bite mark is greater than the number of pixels continuously distributed outward, which means that the corresponding area is unfolded into a sheet along the surface. Combining the relationship between the corresponding area and the boundary of the old bite mark, the old tooth mark rolled-up aluminum sheet can be extracted from the old bite mark.

[0080] The edge where the old toothed rolled aluminum sheet meets the component frame is extracted to obtain the rolled connection edge.

[0081] The edge where the old toothed rolled aluminum sheet meets the module frame is the root boundary where the old toothed rolled aluminum sheet starts to curl up from the surface of the unrolled module frame. One side of the edge belongs to the old toothed rolled aluminum sheet, and the other side belongs to the module frame position that has not been extracted as the old toothed rolled aluminum sheet.

[0082] Specifically, the upper, lower, left, and right sides of the boundary pixels of the old toothed rolled aluminum sheet, as well as the four diagonally adjacent positions, are checked one by one along the outer periphery of the old toothed rolled aluminum sheet. The boundary pixels of the old toothed rolled aluminum sheet that belong to the component frame position but do not belong to the old toothed rolled aluminum sheet are identified as candidate pixels for connecting edges. Adjacent candidate pixels for connecting edges are connected into candidate segments for connecting edges according to the arrangement order of the outer periphery of the old toothed rolled aluminum sheet. The number of candidate pixels for connecting edges in each candidate segment is counted. The candidate segment for connecting edges with the most candidate pixels for connecting edges that is continuously adjacent to the outer boundary of the old bite tooth mark is retained. When the number of candidate pixels for connecting edges in two candidate segments for connecting edges is equal, the candidate segment for connecting edges that is arranged earlier along the outer boundary of the old bite tooth mark is retained. The candidate pixels for connecting edges in the retained candidate segments for connecting edges are arranged in adjacent order to obtain the rolled connecting edge.

[0083] The candidate pixels of the connecting edge are adjacent to both the old toothed rolled aluminum sheet and the frame of the unrolled component. Therefore, they are located at the junction of the rolled material and the surface of the component frame. Connecting the candidate pixels of the connecting edge according to the outer periphery of the old toothed rolled aluminum sheet can form a continuous root boundary. Selecting the connecting edge candidate segment with the largest number of connecting edge candidate pixels and that is continuously adjacent to the outer boundary of the old bite tooth can eliminate the free edge of the outer periphery of the old toothed rolled aluminum sheet and obtain the rolled connecting edge.

[0084] S3. Accumulate the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, obtain the pre-rolling occupied area;

[0085] In an embodiment of the present invention, obtaining the material extension length of each rolled material strip includes:

[0086] The material extension length of the rolled material strip is the actual surface path length of the old toothed rolled aluminum sheet extending from the rolled connection edge to the outer edge of the rolled along the material spreading direction. This length is expressed in the calibrated length unit of the spatial coordinates of the grounding part. Each rolled material strip is associated with a material extension length, which is used to limit the distance that the corresponding rolled material strip covers the surface of the component frame when it is spread in the reverse direction.

[0087] Based on the rolled-up connection edge, the edge of the old toothed rolled-up aluminum sheet is extracted relative to the edge to obtain the rolled-up outer edge;

[0088] Relative edge extraction is a process that uses the rolled-up connecting edge as the fixed side boundary of the old toothed rolled-up aluminum sheet, and determines the other side boundary from the remaining boundaries of the old toothed rolled-up aluminum sheet according to the interval relationship through the sheet material; the rolled-up outer edge is a continuous free boundary in the outer periphery of the old toothed rolled-up aluminum sheet that is located on both sides of the sheet material and is not connected to the component frame position. The rolled-up outer edge is represented by the ordered pixel coordinates in the grounding correction image.

[0089] Specifically, pixels whose at least one adjacent position in the old toothed rolled aluminum sheet does not belong to the old toothed rolled aluminum sheet are identified as boundary pixels of the old toothed rolled aluminum sheet. Pixels belonging to the rolling connection edge are removed from the boundary pixels of the old toothed rolled aluminum sheet to form outer edge candidate pixels. Adjacent outer edge candidate pixels are connected according to the arrangement order of the outer periphery of the old toothed rolled aluminum sheet to form outer edge candidate segments. Starting from each outer edge candidate pixel, the process expands layer by layer inside the old toothed rolled aluminum sheet according to the top, bottom, left, right, and four diagonally adjacent positions. The number of adjacent positions traversed when first reaching the rolling connection edge is determined as the number of connection edge interval layers for the corresponding outer edge candidate pixel. All outer edge candidate pixels within each outer edge candidate segment are then... The number of connecting edge spacing layers is added together and divided by the number of outer edge candidate pixels contained in the corresponding outer edge candidate segment to form the average spacing layer of the outer edge candidate segment. When the number of outer edge candidate pixels contained in the outer edge candidate segment is 0, the corresponding outer edge candidate segment is removed. The outer edge candidate segment with the largest average spacing layer is selected. When the average spacing layer of multiple outer edge candidate segments is equal, the outer edge candidate segment with the largest number of outer edge candidate pixels is selected. When the number of outer edge candidate pixels is still equal, the outer edge candidate segment with the outer edge candidate pixels arranged in the order of the old tooth marks rolled around the outer periphery of the aluminum sheet is selected. The outer edge candidate pixels in the selected outer edge candidate segment are arranged in adjacent order to obtain the rolled outer edge.

[0090] The roll-up connecting edge defines the fixed side where the old toothed roll-up aluminum sheet connects to the component frame. The number of connecting edge layers indicates the range of sheet material spanned from the other boundaries to the fixed side inside the old toothed roll-up aluminum sheet. The continuous boundary with the maximum average number of outer edge candidate segments is located on the opposite side of the fixed side. Combining the number of outer edge candidate pixels and the outer perimeter arrangement order results in a unique and continuous roll-up outer edge.

[0091] Based on the connecting edge and the outer edge of the roll, the old toothed rolled aluminum sheet is divided into strips to obtain the rolled material strips;

[0092] Strip division is the process of dividing the old toothed rolled aluminum sheet into adjacent sheet-like areas along the material spreading direction from the rolled connecting edge to the rolled outer edge. Each sheet-like area has a corresponding boundary position on the rolled connecting edge and the rolled outer edge. The rolled material strip is a continuous aluminum material area formed by the strip division. Each rolled material strip covers the old toothed rolled aluminum sheet and is adjacent to each other in sequence along the arrangement direction of the rolled connecting edge.

[0093] Specifically, based on the ordered pixel coordinates of the flipped connecting edge and the flipped outer edge, the pixels within them are assigned connecting edge position indices and outer edge position indices starting from 0 and incrementing. Each connecting edge position indice is divided by the number of flipped connecting edge pixels minus 1 to form the relative position of the connecting edge. Similarly, each outer edge position indice is divided by the number of flipped outer edge pixels minus 1 to form the relative position of the outer edge. When the number of flipped connecting edge pixels or the number of flipped outer edge pixels is 1, the corresponding relative position of the connecting edge or the relative position of the outer edge is set to 0. For each relative position of the connecting edge, the outer edge relative position with the smallest absolute value of the subtraction result is selected, and the corresponding flipped connecting edge pixels and flipped outer edge pixels are combined to form the corresponding position of the outer edge. When the absolute values ​​of the subtraction results of multiple outer edge relative positions are equal, the outer edge relative position with the smallest outer edge position indice is selected. Starting from the pixel of the rolled-up connecting edge at the corresponding position of each connecting edge, adjacent pixels are connected layer by layer inside the old toothed rolled aluminum sheet until the corresponding rolled-up outer edge pixel is reached. The connecting path with the minimum number of adjacent pixels is determined as the strip boundary path. When there are multiple connecting paths with the same number of adjacent pixels, the connecting path with the first order is selected according to the arrangement of the vertical and horizontal coordinates of the pixels in the path from small to large. The pixels belonging to the old toothed rolled aluminum sheet between two adjacent strip boundary paths are combined with the two strip boundary paths to form a strip pixel region according to the ascending order of the connecting edge position number. When there are fewer than 2 strip boundary paths, the process of extracting the rolled-up connecting edge and rolled-up outer edge is returned. All strip pixel regions are arranged in ascending order of the connecting edge position number to obtain the rolled material strip.

[0094] The relative positions of the connecting edges and the outer edges unify the rolling connecting edges and rolling outer edges with different pixel counts to an arrangement reference from 0 to 1. The position with the smallest absolute value of the subtraction result represents the corresponding position on the two edges along their respective arrangement directions. The boundary between adjacent material parts is defined by the shortest connection path inside the old toothed rolling aluminum sheet between the corresponding positions of the connecting outer edges. The pixels between the boundary paths of adjacent strips are sequentially collected to cover the old toothed rolling aluminum sheet and obtain the rolling material strip.

[0095] Based on the arrangement position of the rolled material strips on the rolled connection edge, the rolled material strips are arranged sequentially to obtain a rolled material strip sequence.

[0096] The arrangement of the rolled material strips on the rolled connection edge is the range of the index of the pixels where each rolled material strip intersects with the rolled connection edge in the ordered pixel coordinates of the rolled connection edge; the sequential arrangement is the process of organizing the rolled material strips from the start end to the end end of the rolled connection edge according to each index range; the rolled material strip sequence is an ordered set that retains all rolled material strips and their arrangement order.

[0097] Specifically, for each rolled material strip, extract the strip connecting pixels that simultaneously belong to both the rolled material strip and the rolled connecting edge. Obtain the connecting edge position number corresponding to each strip connecting pixel from the ordered pixel coordinates of the rolled connecting edge. Determine the minimum value among all the connecting edge position numbers of the same rolled material strip as the starting position number of the corresponding rolled material strip arrangement and the maximum value as the ending position number of the corresponding rolled material strip arrangement. When any rolled material strip does not have any strip connecting pixels, return the corresponding rolled material strip to the strip division process. Arrange all rolled material strips in ascending order of the starting position number. When the starting position numbers of two rolled material strips are equal, arrange them in ascending order of the ending position number. When the ending position numbers are still equal, arrange them in the order of formation of the rolled material strips in the strip division process. Combine all the arranged rolled material strips with their starting and ending position numbers to form a rolled material strip sequence.

[0098] The strip connection pixels represent the actual contact range of the rolled material strips on the rolled connection edge. Arranging the start position number and the end position number converts the contact range into an interval under the same rolled connection edge coordinate reference. By comparing the start point of the interval, the end point of the interval, and the strip formation order in sequence, sorting conflicts caused by the same intersection range can be eliminated and a rolled material strip sequence with the extension direction of the rolled connection edge can be formed.

[0099] Along the extension path from the connecting edge to the outer edge of each rolled material strip in the sequence of rolled material strips, obtain the spatial coordinates of the corresponding position of the rolled material strip;

[0100] The extension path is a continuous pixel path inside each rolled material strip, passing through the sheet material from one side of the rolled connection edge to the other side of the rolled edge. The extension path is determined one by one according to the sequence of rolled material strips. The spatial coordinates of the corresponding position of the rolled material strip are the three-dimensional positions of the entity points represented by each pixel in the extension path in the calibration coordinate system of the image acquisition device. The spatial coordinates use the same calibration length unit with the same spatial depth information and are arranged according to the order of the extension path.

[0101] Specifically, the rolled material strips are processed one by one according to their sequence. From the pixels where each rolled material strip intersects with the rolling connection edge, the pixel with the middle position number of the connection edge is selected as the starting pixel for material extension. Similarly, from the pixels where each rolled material strip intersects with the outer edge, the pixel with the middle position number of the outer edge is selected as the ending pixel for material extension. When the number of pixels is even, the pixel with the smallest position number among the two middle pixels is selected. Starting from the starting pixel, adjacent pixels are connected layer by layer within the corresponding rolled material strip until the ending pixel is reached. The connection path with the smallest number of adjacent pixels is determined as the material extension path. When multiple connection paths with the same number of adjacent pixels exist, the connection path with the earlier pixel coordinates is selected. The corresponding position of the registration reference image is obtained from the pixel source correspondence of each shooting direction retained in the perspective correction image of the grounding part according to the pixel order of the material extension path. The imaging center coordinates, imaging focal length, and pixel size of the registration reference image are obtained from the calibration data of the image acquisition device. The process involves subtracting the x-coordinate of the corresponding pixel in the reference image from the x-coordinate of the image center, multiplying by the pixel size, and dividing by the imaging focal length to form the lateral line-of-sight component. Similarly, subtracting the y-coordinate of the corresponding pixel in the registration reference image from the y-coordinate of the image center, multiplying by the pixel size, and dividing by the imaging focal length to form the longitudinal line-of-sight component. The lateral line-of-sight component, the longitudinal line-of-sight component, and the imaging direction vector of the registration reference image are combined according to the calibration coordinates of the image acquisition device and divided by the square root of the sum of the squares of each component to form the unit line-of-sight direction. Each coordinate component of the unit line-of-sight direction is multiplied by the spatial depth information of the corresponding pixel in the grounding correction image and added to the corresponding component of the imaging center coordinates of the registration reference image to form the spatial coordinates of the material extension path position. When any pixel cannot obtain the corresponding position or spatial depth information of the registration reference image, the corresponding rolled-up material strip is returned to the grounding perspective correction image to obtain the spatial depth information of each corresponding position. The spatial coordinates of each material extension path position are arranged according to the pixel order of the material extension path and aggregated according to the rolled-up material strip sequence to obtain the spatial coordinates of the corresponding position of the rolled-up material strip.

[0102] The starting and ending pixels of the material extension are located on the fixed and free sides of the rolled material strip, respectively. The material extension path inside the rolled material strip with the minimum number of adjacent pixels represents the continuous material spreading direction between the two sides. The corresponding position of the registration reference image and the calibration data form the path unit line of sight from the imaging center to the corresponding entity point. The path unit line of sight and spatial depth information are combined to determine the spatial coordinates of the material extension path position of the entity point in the same calibration coordinate system. After being collected according to the path and strip order, the spatial coordinates of the corresponding position of the rolled material strip are formed.

[0103] Based on the spatial coordinates of the corresponding positions of the rolled material strips, the spatial distances between adjacent positions in each rolled material strip are accumulated to obtain the material extension length of each rolled material strip.

[0104] The spatial distance between adjacent positions is the straight-line interval between two adjacent entity points in the calibration coordinate system along the material extension path of the same rolled material strip; the cumulative spatial distance is the sum of the spatial distances between all adjacent positions in the order of the material extension path, in order to approximately represent the path length of the rolled material strip extending along the curved surface from the rolled connection edge to the rolled outer edge.

[0105] Specifically, spatial coordinates of the corresponding positions of the rolled material strips are selected one by one according to the sequence of rolled material strips. For two adjacent material extension path positions in the same rolled material strip according to the material extension path order, the horizontal, vertical, and depth coordinate components of the latter material extension path position are subtracted from the corresponding coordinate components of the former material extension path position to form the horizontal coordinate difference, vertical coordinate difference, and depth coordinate difference of adjacent positions. The squares of the horizontal coordinate difference, vertical coordinate difference, and depth coordinate difference of adjacent positions are added together and the square root is taken to form the spatial distance between adjacent positions. The spatial distances of all adjacent positions of the same rolled material strip are added together according to the material extension path order to form the material extension length of the corresponding rolled material strip. When the same rolled material strip contains less than two material extension path position spatial coordinates, the corresponding rolled material strip is returned to the process of obtaining the spatial coordinates of the corresponding position of the rolled material strip. The material extension lengths of all rolled material strips are collected according to the arrangement order of the rolled material strip sequence to obtain the material extension length of each rolled material strip.

[0106] The difference in lateral coordinates, the difference in longitudinal coordinates, and the difference in depth coordinates of adjacent positions together represent the positional changes of adjacent entity points in the three-dimensional calibration coordinate system on the material extension path. The square root of the sum of the squares of the three coordinate differences forms the spatial distance between adjacent positions. By accumulating all the spatial distances between adjacent positions in the order from the rolled-up connecting edge to the rolled-up outer edge of the material extension path, the curvature path of the rolled-up material strip can be preserved and the material extension length of each rolled-up material strip can be obtained.

[0107] In an embodiment of the present invention, the pre-rollover occupancy area is obtained, including:

[0108] The pre-rollover occupancy area is the area on the component frame surface and covered by the corresponding aluminum material before the old toothed aluminum sheet is rolled over. This area is formed by splicing together each rolled material strip with the joint position on the rolling connection edge as a fixed starting point, and spreading in the opposite direction of the current rolling direction according to the extension length of each material. It is used to indicate the planar occupancy range of the corresponding component frame material before the old toothed marks are formed.

[0109] The position of each rolled material strip on the rolled connection edge is taken as the starting point for the spread of the rolled material strip;

[0110] The spreading start point is the set of strip connection pixels where each rolled material strip intersects with the rolled connection edge. The spreading start point is located at the root of the material where the rolled material strip is still connected to the component border position, and is associated with the corresponding start position number and end position number according to the rolled material strip sequence, which is used to maintain the fixed side position and mutual order of each rolled material strip during the reverse spreading process.

[0111] Specifically, roll material strips are selected one by one according to the roll material strip sequence. From the corresponding roll material strip, strip connection pixels that are both roll connection edges and whose connection edge position numbers are between the start and end positions of the corresponding arrangement are retained. The retained strip connection pixels are arranged according to the ordered pixel coordinates of the roll connection edges. When any roll material strip does not have any retained strip connection pixels, the corresponding roll material strip is returned to the strip division process. All the retained strip connection pixels corresponding to each roll material strip and their arrangement order are determined as the starting point for the roll material strip spreading.

[0112] Based on the starting point and the length of the material extension, each rolled material strip in the rolled material strip sequence is rolled in reverse to obtain the restored area;

[0113] Reverse spreading involves keeping the spreading starting point stationary and spreading each rolled material strip along a side perpendicular to the local rolled connection edge and away from the current position of the old toothed rolled aluminum sheet until the spreading path reaches the corresponding material extension length. The restoration area is the set of pixel areas covered on the component frame position after each rolled material strip is reverse spread, and each restoration area corresponds to a rolled material strip and its spreading starting point.

[0114] Specifically, the spreading starting point of each of the rolled material strips is processed sequentially according to the roll-up material strip sequence. The local connection edge direction is obtained from the preceding and following pixels of each spreading starting point pixel in the roll-up connection edge arrangement. When the spreading starting point pixel is located at the endpoint of a roll-up connection edge, the direction between the adjacent roll-up connection edge pixel and the spreading starting point pixel is used. The horizontal and vertical components of the local connection edge direction are squared respectively, summed, and the square root is taken to form the local connection edge direction length. The horizontal and vertical components of the local connection edge direction are divided by the local connection edge direction length to form the local connection edge unit direction. When the local connection edge squared... When the length is 0, return to the rolling connection edge extraction process corresponding to the spreading start point. Take the inverse of the vertical component of the local connection edge in the unit direction and form the first vertical direction with the horizontal component. Take the inverse of the horizontal component of the local connection edge in the unit direction and form the second vertical direction with the vertical component. Check whether the pixels adjacent to the spreading start point in the first and second vertical directions belong to the corresponding rolling material strip. Determine the vertical direction of the adjacent pixels belonging to the corresponding rolling material strip as the current rolling direction and determine the other vertical direction as the reverse spreading direction. When the adjacent pixels in both vertical directions do not belong to the corresponding rolling material strip, obtain the two adjacent pixels respectively. The minimum pixel distance to the corresponding rolled material strip is used to determine the current rolling direction, with the vertical direction being the minimum pixel distance. When two minimum pixel distances are equal, the current rolling direction is selected according to the order of the horizontal component and then the vertical component from smallest to largest. Starting from each starting pixel, the rolling direction extends pixel by pixel along the reverse rolling direction within the component border. The spatial coordinates of each extended pixel in the ground correction image are compared with the spatial coordinates of the previous extended pixel using the square root of the sum of the squares of the differences in the horizontal, vertical, and depth coordinates to form the spatial distance between adjacent rolling positions. The spatial distances between adjacent rolling positions are then calculated sequentially according to the extension order. The cumulative spreading distance is increased. When the cumulative spreading distance is less than the material extension length of the corresponding rolled material strip, the spreading continues and stops when the cumulative spreading distance is greater than or equal to the material extension length of the corresponding rolled material strip. When the cumulative spreading distance is still less than the material extension length of the corresponding rolled material strip when the component edge position boundary is reached, the multi-view image of the corresponding grounding part is returned to the multi-directional image acquisition process to expand the acquisition range of the grounding part. All extended pixels at the same spreading starting point are connected according to the adjacent relationship to form strip complex atomic regions. The strip complex atomic regions of each rolled material strip are collected according to the rolled material strip sequence to obtain the restored region.

[0115] The local connection edge unit direction defines the extension direction of the fixed side of the rolled material strip. Two directions perpendicular to the local connection edge unit direction are located on both sides of the rolled connection edge. The current rolling direction is determined according to the current position of the corresponding rolled material strip, and the opposite side can be taken to form a reverse spreading direction. The cumulative spreading distance is the three-dimensional spatial distance between adjacent entity points on the component frame position and the corresponding material extension length is used as the stopping condition, so that each strip complex atomic region maintains the actual extension scale of the rolled material strip and forms a restoration region.

[0116] Following the arrangement order of the rolled-up material strips, the restoration area is sequentially spliced ​​together to obtain the pre-rolled-up occupant area.

[0117] Sequential splicing involves merging the restoration areas corresponding to adjacent rolled material strips in sequence according to the rolled material strip sequence, and retaining only one coordinate position for repeated pixels; the pre-rolling occupancy area is the continuous component border surface area after sequential splicing, and the side closest to the rolled connection edge maintains the original arrangement relationship of each spreading starting point, and its outward extension range is limited by the material extension length of each rolled material strip.

[0118] Specifically, the complex atomic region corresponding to the first rolled material strip in the sequence is determined as the spliced ​​and restored region. Following the sequence, the complex atomic regions corresponding to the next rolled material strip are selected sequentially. It is checked whether the current complex atomic region and the spliced ​​and restored region contain pixels with the same coordinates, or pixels that are adjacent to each other on the top, bottom, left, and right sides. If pixels with the same coordinates exist, the duplicate pixels are removed from the current complex atomic region, and the remaining pixels are merged into the spliced ​​and restored region. If no pixels with the same coordinates exist, but adjacent pixels exist, all pixels in the current complex atomic region are merged into the spliced ​​and restored region. If neither pixels with the same coordinates nor adjacent pixels exist, the current complex atomic region is returned to the reverse spreading process to redetermine the reverse spreading direction until the current complex atomic region can be merged with the spliced ​​and restored region. The merging of all complex atomic regions is completed according to the rolled material strip sequence. The merged spliced ​​and restored regions are arranged according to the pixel coordinates in the grounding correction image to obtain the pre-rolling occupancy region.

[0119] The sequence of rolled material strips maintains the original adjacent relationship of each rolled material strip on the rolling connection edge. Merging the complex atomic regions of the strips in this order can maintain the lateral arrangement of the aluminum material before rolling. Removing duplicate pixels with the same coordinates can avoid overlapping positions being repeatedly occupied. Using coordinate coincidence or adjacency as the continuity condition and returning the reverse spreading process to discontinuous areas can connect all the complex atomic regions of the strips into the pre-rolling occupied area.

[0120] S4. Based on the area occupied before the rollover, the current occlusal tooth area is extracted by regional overlap to obtain the cross-old trace occlusal bearing area.

[0121] In an embodiment of the present invention, a cross-marking engagement area is obtained, including:

[0122] The cross-old-mark biting and bearing area is the component frame surface range within the current grounding pad's biting tooth action range that simultaneously falls into the old tooth mark backfill area and the pre-rollover occupying area. The cross-old-mark biting and bearing area is represented by pixel coordinates in the grounding part correction image and is used to determine the actual position where the current biting tooth crosses the old biting tooth mark and is borne by the composite material.

[0123] Connect the current position of each rolled material strip in the rolled material strip sequence with the corresponding restoration area to obtain the material return path;

[0124] The current position of the rolled material strip is the pixel area occupied by the rolled material strip when it is in the rolled state in the old toothed rolled aluminum sheet. This pixel area is associated with the strip identifier and arrangement order in the rolled material strip sequence. The material return path is an ordered coordinate sequence extending from the current position pixel of the rolled material strip to the corresponding strip complex atomic region pixel in the restoration area. The material return path represents the corresponding direction between each position when the rolled material strip returns from the rolled state to the spread state.

[0125] Specifically, the rolled material strips are selected one by one according to the rolled material strip sequence. The starting point of the corresponding rolled material strip is taken as layer 0. In the current position of the rolled material strip, the strip is expanded layer by layer according to the top, bottom, left, right and four diagonally adjacent positions. The layer number to which the pixel at the current position is first expanded is determined as the extension layer number of the pixel at the current position. All pixels at the current position are arranged in ascending order of extension layer number, vertical coordinate and horizontal coordinate to form the pixel sequence of the current position. The strips in the restoration area corresponding to the same starting point as layer 0 are then identified. Within the complex atomic region, the layers are expanded layer by layer according to the same adjacency relationship. The layer number to which the pixel of the complex atomic region is first expanded is determined as the extension layer number of the restored position pixel. All pixels of the complex atomic region are arranged in ascending order of the restored position pixel extension layer number, ordinate, and abscissa to form the restored position pixel sequence. When the current position pixel sequence or the restored position pixel sequence does not contain any pixels, the corresponding rolled-up material strip is returned to the reverse spreading process. For each current position pixel index starting from 0 in the current position pixel sequence, the current position pixel index is multiplied by the restored position image. The length of the pixel sequence is reduced by 1 and divided by the length of the current position pixel sequence reduced by 1. The result of the division is rounded to form the corresponding restoration index. When the length of the current position pixel sequence is 1, the restoration index is set to 0. The corresponding restored position pixel is selected from the restored position pixel sequence according to the restoration index. The maximum value between the absolute value of the difference between the x-coordinate of the current position pixel and the x-coordinate of the restored position pixel and the absolute value of the difference between the y-coordinate of the current position pixel and the y-coordinate of the restored position pixel is determined as the number of back coordinate steps. When the number of back coordinate steps is greater than 0, the x-coordinate of the current position pixel is set to 0. Add the corresponding coordinate difference to the ordinate and multiply it by the current step number from 0 to the number of return coordinate steps, and divide by the number of return coordinate steps. Round the result of each addition to form the return path pixel coordinates. When the number of return coordinate steps is 0, the current position pixel coordinates are determined as the return path pixel coordinates. Arrange the return path pixel coordinates between the same current position pixel and the corresponding restored position pixel according to the current step number to form the material return sub-path. Collect all the material return sub-paths of the rolled material strips according to the rolled material strip sequence and the current position pixel sequence to obtain the material return path.

[0126] The current position pixel extension layer number and the restored position pixel extension layer number unify the rolled material strip and the corresponding strip complex atomic region into an arrangement reference extending outward from the spreading start point. The restored corresponding sequence number associates each current position pixel of the rolled material strip with the corresponding spreading position according to the length ratio of the two pixel sequences. The number of back-shift coordinate steps and step-by-step coordinate interpolation connect the associated positions into a continuous coordinate sequence. Therefore, the rolling position and spreading position of each rolled material strip can form a material back-shift path.

[0127] Based on the material return path, the current position of each rolled material strip is mapped to the corresponding restoration area to obtain the old tooth mark backfill area;

[0128] Mapping is the process of establishing a coordinate correspondence between the current position pixel of the rolled material strip and the restored position pixel at the end of the path according to each material return sub-path; the old tooth mark backfill area is the component frame surface area that is jointly covered by each rolled material strip after returning to the corresponding strip's original atomic area along the material return path. The old tooth mark backfill area is located in the pre-rolling occupied area and is represented by the pixel coordinates in the grounding correction image.

[0129] Specifically, the material return path is processed line by line according to the sequence of rolled material strips and the sequence of pixels at the current position. From each material return sub-path, the current position pixel at the starting point and the restored position pixel at the ending point are obtained. When the restored position pixel belongs to the complex atomic region of the rolled material strip corresponding to the current position pixel in the restored region, a return mapping relationship is established between the current position pixel and the restored position pixel, and the restored position pixel is determined as the backfill landing point pixel. When the restored position pixel does not belong to the corresponding complex atomic region, the corresponding material return sub-path is returned to the material return path formation process. For each pixel in the same complex atomic region that is not determined as a backfill landing point pixel, the pixel's horizontal and vertical coordinates are respectively mapped to each backfill landing point in that complex atomic region. Subtract the corresponding coordinates of the pixels, add the squares of the two coordinate differences, and take the square root to form the backfill landing pixel distance. Select the backfill landing pixel with the smallest backfill landing pixel distance and extend the corresponding backshift mapping relationship to the pixels that have not been determined as backfill landing pixels. When the backfill landing pixel distances of multiple backfill landing pixels are equal, select the backfill landing pixel with the smallest corresponding sequence number to restore. When there are no backfill landing pixels in the corresponding strip complex atomic region, return the corresponding rolled material strip to the material backshift path formation process. Connect all restored position pixels with backshift mapping relationship and all pixels that have not been determined as backfill landing pixels according to the adjacency relationship of the strip complex atomic region, and merge them according to the rolled material strip sequence to obtain the old tooth mark backfill area.

[0130] The starting and ending points of the material return path represent the material positions of the same rolled material strip in the rolled and spread states, respectively. The return mapping relationship allows the current pixel to return to the corresponding strip complex atom region. The association with the minimum pixel distance of the backfill landing point assigns the pixels not covered by the discrete mapping to the spatially adjacent material positions. After connecting and merging the strip complex atom regions, the old tooth mark backfill area of ​​the old bite mark position of each rolled material strip can be formed.

[0131] Get the current occlusal tooth region;

[0132] The current biting tooth area is the continuous pixel area covered by each biting tooth pressed into the component frame surface in the current installation state of the grounding pad. The current biting tooth area is defined by the biting recess area and biting center position in the corresponding biting position of the grounding pad, and is located within the biting contact range enclosed by the edge of the grounding pad.

[0133] Specifically, the occlusal recess area and occlusal center position are obtained one by one from the occlusal positions corresponding to the grounding pad. Each occlusal recess area is divided into occlusal recess connected regions according to the upper, lower, left, right, and four diagonally adjacent positions. The occlusal recess connected regions containing the corresponding occlusal center position are determined as the current occlusal tooth candidate regions. When no occlusal recess connected region contains the corresponding occlusal center position, the horizontal and vertical coordinates of the corresponding occlusal center position are subtracted from the corresponding coordinates of each pixel in each occlusal recess connected region. The squares of the two coordinate differences are added together and the square root is taken to form the pixel distance of the center position. The occlusal tooth containing the pixel with the smallest pixel distance of the center position is selected. The concave connected region is used as the current bite tooth candidate region. When the center pixel distance of multiple pixels is equal, the bite concave connected region containing the pixel with the smallest vertical coordinate and the smallest horizontal coordinate when the vertical coordinates are equal is selected. The horizontal coordinates of the intersection of the current bite tooth candidate region boundary and the same pixel row are obtained row by row. The horizontal coordinates are arranged from small to large and the pixels between two adjacent horizontal coordinates and located within the edge of the grounding pad are included in the current bite tooth candidate region. When the number of intersecting horizontal coordinates is odd, the horizontal coordinates that are missing pairing at the end of the arrangement are discarded. All current bite tooth candidate regions corresponding to the bite center position are merged according to the pixel coordinates of the grounding part corrected image to obtain the current bite tooth region.

[0134] The bite center position is located at the local recessed center formed by the bite teeth of the grounding pad. The bite center position or the bite recessed connected region with the minimum center position pixel distance to the bite center position corresponds to the current bite tooth pressing range. By filling the boundary intersections in pairs, the closed recessed boundary can be converted into a complete coverage area. After merging all the regions corresponding to the bite center positions, the current bite tooth region is formed.

[0135] Based on the current occlusal tooth area, the old tooth mark backfill area, and the area occupied before rollover, the current occlusal tooth area is extracted by regional overlap to obtain the occlusal bearing area across the old tooth mark.

[0136] Region overlap extraction involves checking whether the pixels in the current biting tooth region belong to both the old tooth mark backfill area and the pre-rollover occupied area under the same pixel coordinate reference of the grounding part correction image, and retaining the pixels shared by all three. The retained pixels represent the actual position of the current biting tooth that was originally supported by the old tooth mark backfill material.

[0137] Specifically, the pixel coordinates of the grounding correction image are uniformly adopted for the current occlusal tooth area, the old tooth mark backfill area, and the pre-rollover occupancy area. The horizontal and vertical coordinates of the pixels in the current occlusal tooth area are checked one by one to see if they are equal to the horizontal and vertical coordinates of the pixels in the old tooth mark backfill area and the pre-rollover occupancy area. The pixels in the current occlusal tooth area that exist in both the old tooth mark backfill area and the pre-rollover occupancy area are identified as triple overlapping pixels. The triple overlapping pixels are connected into triple overlapping connected areas according to the upper, lower, left, and right sides and four diagonal adjacent relationships. The triple overlapping connected area pixels located outside the edge of the grounding pad are removed. All the retained triple overlapping connected area pixels are arranged according to the coordinates in the current occlusal tooth area. When there are no triple overlapping pixels, the area record without pixels is retained to obtain the cross-old tooth mark occlusal bearing area.

[0138] The current biting tooth area defines the actual range of action of the current biting tooth, the old tooth mark backfill area defines the range occupied by the rolled material after returning to the old biting tooth mark, and the pre-rollover occupancy area defines the range of the component frame surface covered by the rolled material before the rollover occurs. Pixels with the same coordinates of the three simultaneously satisfy the current biting action and the old tooth mark material bearing relationship. Connecting and retaining these pixels can obtain the biting bearing area across the old tooth mark.

[0139] S5. Extract the area closest to the current biting tooth area to obtain the target bearing area; based on the cross-old mark biting bearing area and the target bearing area, monitor and maintain the grounding pad of the target photovoltaic module.

[0140] In an embodiment of the present invention, the target receiving area is obtained, including:

[0141] The target receiving area is a continuous area on the component frame that avoids the old bite marks and the area occupied before the roll-up, can completely accommodate the outer contour of the current bite area, and has the minimum spatial distance from the current bite area. The target receiving area is represented by the pixel coordinates in the grounding correction image and is used to limit the position of the component frame surface where the bite teeth can be pressed in when the grounding pad is reinstalled.

[0142] Remove the old bite marks and the pre-rollover area from the component border position to obtain the remaining border area;

[0143] The remaining border area is a set of pixels in the component border position that do not belong to the old bite marks and do not belong to the pre-rollover occupied area. The remaining border area retains the component border surface that was not occupied by the original occupied area of ​​the historical bite recess and the rolled-up material, and uses the pixel coordinates and spatial depth information of the grounding part to correct the image.

[0144] Specifically, the pixel coordinates of the component border position, old bite marks, and pre-rollover occupied area are uniformly corrected using the pixel coordinates of the ground part image. The component border position pixel is checked one by one to see if it exists in the old bite marks or pre-rollover occupied area. When the horizontal and vertical coordinates of the component border position pixel are equal to the horizontal and vertical coordinates of the old bite mark pixel or the pre-rollover occupied area pixel, the corresponding component border position pixel is removed. When the component border position pixel is not equal to the coordinates of the old bite mark pixel or the pre-rollover occupied area pixel, the corresponding component border position pixel is retained. The retained component border position pixels are connected according to the top, bottom, left, right and four diagonal adjacent relationships, and the spatial depth information associated with each pixel is retained. When no component border position pixels are retained, a border area without pixels is formed, and the remaining border area is obtained.

[0145] Old bite marks indicate the component frame surface where historical bite indentation has occurred. The pre-rollover occupied area indicates the component frame surface occupied before the old bite marks and rolled aluminum sheet rolled over. By removing the two types of ranges from the component frame position according to the same pixel coordinates, the remaining pixels correspond only to the component frame surface not occupied by old bite marks and rolled-over material, thus forming the remaining frame area.

[0146] Using the outer contour of the current biting tooth area, perform morphological erosion on the remaining border area to obtain the area that the biting tooth can accommodate;

[0147] The outer contour of the current biting tooth region is an ordered set of boundary pixels in the current biting tooth region whose at least one adjacent position does not belong to the current biting tooth region; morphological erosion is the process of translating the outer contour of the current biting tooth region relative to the reference pixel to each candidate pixel in the remaining border region, and retaining only the candidate pixels whose outer contour pixels are still located in the remaining border region after translation; the biting tooth accommodating region is the region composed of all candidate pixels that meet the complete contour accommodating condition.

[0148] Specifically, from the current occlusal tooth region, at least one pixel on the upper, lower, left, and right sides, as well as four diagonally adjacent pixels that do not belong to the current occlusal tooth region, are extracted to form the outer contour of the current occlusal tooth region. The x-coordinates of all pixels in the current occlusal tooth region are summed and divided by the number of pixels in the current occlusal tooth region to form the x-coordinate of the occlusal tooth center. The y-coordinates of all pixels in the current occlusal tooth region are summed and divided by the number of pixels in the current occlusal tooth region to form the y-coordinate of the occlusal tooth center. When the number of pixels in the current occlusal tooth region is 0, the process of obtaining the current occlusal tooth region is returned. The absolute value of the difference between the x-coordinate and the x-coordinate of each pixel in the current occlusal tooth region and the absolute value of the difference between the y-coordinate and the y-coordinate of each pixel in the current occlusal tooth region are summed to form the center coordinate interval. The pixel in the current occlusal tooth region with the smallest center coordinate interval is selected as the reference pixel for the occlusal tooth contour. When the center of multiple pixels in the current occlusal tooth region... When the coordinate intervals are equal, select the pixel in the current biting tooth region that has the smallest vertical coordinate and the smallest horizontal coordinate when the vertical coordinates are equal. Subtract the corresponding coordinates of the reference pixels of the biting tooth contour from the horizontal and vertical coordinates of each pixel in the outer contour of the current biting tooth region to form the horizontal and vertical offsets of the contour. Select the pixels in the remaining border region one by one as erosion candidate pixels. Add the horizontal and vertical offsets of the contour to the horizontal and vertical coordinates of the erosion candidate pixels to form the translation contour pixel coordinates. When all translation contour pixel coordinates belong to the remaining border region, retain the corresponding erosion candidate pixels. When any translation contour pixel coordinate does not belong to the remaining border region, remove the corresponding erosion candidate pixels. Connect all the retained erosion candidate pixels according to the adjacent relationship. When no erosion candidate pixels are retained, a non-pixel-containing area is formed, and the biting tooth can be contained area is obtained.

[0149] The reference pixels of the bite tooth contour and each contour offset convert the outer contour of the current bite tooth region into contour coordinates that can be translated point by point in the remaining border region. Only when all translated contour pixel coordinates belong to the erosion candidate pixels of the remaining border region can the outer contour of the current bite tooth region avoid the old bite tooth marks and the pre-rollover occupied area. Retaining these erosion candidate pixels can form the bite tooth accommodating area.

[0150] Extract the region closest to the current occlusal tooth region from the region that can accommodate the occlusal teeth to obtain the target receiving region.

[0151] The region closest to the current occlusal tooth region is the connected region with the minimum straight-line pixel distance between it and any pixel in the current occlusal tooth region. This distance is calculated using the horizontal and vertical coordinates of the grounding correction image and is used to determine the shortest continuous bearing range in a position that satisfies the complete contour bearing condition.

[0152] Specifically, the area that can accommodate the occlusal teeth is divided into accommodating connected regions according to the top, bottom, left, right, and four diagonal adjacency relationships. For each pixel in each accommodating connected region and each pixel in the current occlusal tooth region, the horizontal coordinate difference is obtained by subtracting the horizontal coordinate of the pixel in the accommodating connected region from the horizontal coordinate of the pixel in the current occlusal tooth region, and the vertical coordinate difference is obtained by subtracting the vertical coordinate of the pixel in the accommodating connected region from the vertical coordinate of the pixel in the current occlusal tooth region. The square root of the sum of the squares of the horizontal and vertical coordinate differences is used to obtain the pixel distance between the regions. All region images formed by the same accommodating connected region are selected. The minimum value in the pixel distance is used as the current region spacing of the corresponding connected region. All connected regions are arranged in ascending order of their current region spacing. When the current region spacing of two connected regions is equal, the pixel with the minimum vertical coordinate in the two connected regions is compared and the connected region containing the pixel with the minimum vertical coordinate is selected. When the vertical coordinates are still equal, the connected region containing the pixel with the minimum horizontal coordinate is selected. The connected region with the first position in the arrangement is determined as the target receiving region. When the area that the biting teeth can accommodate does not contain any pixels, the target receiving region without pixels is determined, thus obtaining the target receiving region.

[0153] Each accommodating connected region formed by morphological erosion can completely accommodate the outer contour of the current biting tooth region. The region pixel distance represents the straight-line distance between the accommodating connected region and the current biting tooth region under the same image coordinate reference. The minimum region pixel distance of each accommodating connected region limits the shortest pixel interval required to move from the current biting position to the corresponding accommodating connected region. Selecting the accommodating connected region with the minimum current region spacing can obtain the target receiving region.

[0154] In an embodiment of the present invention, monitoring and maintenance of the grounding pad of the target photovoltaic module includes:

[0155] Based on the interlocking area across the old marks, determine the current interlocking position of the grounding pad in the target photovoltaic module;

[0156] Based on the target area, determine the reinstallation location of the grounding pad in the target photovoltaic module;

[0157] Determine the position adjustment relationship of the grounding pad based on the current engagement position and the reinstallation position;

[0158] By matching the position adjustment relationship of the grounding pads with the target photovoltaic modules, the monitoring and operation and maintenance results of the photovoltaic power station can be obtained;

[0159] Based on the monitoring and maintenance results, the grounding pads of the target photovoltaic modules are monitored and maintained.

[0160] An embodiment of the present invention also provides a monitoring and operation and maintenance system for a photovoltaic power plant.

[0161] In this embodiment, the functions of each module / unit are as follows:

[0162] The calibration module performs perspective correction on the registration image of the grounding part to obtain a perspective-corrected image of the grounding part; it then extracts an image region including the grounding pad, the occlusion position, and the occlusion periphery from the perspective-corrected image of the grounding part to obtain a grounding part calibration image.

[0163] The image correction module, based on the grounding location, obtains the image of the image being flipped and connected.

[0164] The in-situ restoration module accumulates the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, the pre-rolling occupied area is obtained;

[0165] The occlusal reconstruction module extracts the overlapping areas of the current occlusal tooth area based on the pre-rollover occupied area, and obtains the cross-old trace occlusal bearing area.

[0166] The operation and maintenance positioning module extracts the area closest to the current biting tooth area to obtain the target bearing area; based on the biting bearing area across old marks and the target bearing area, it monitors and maintains the grounding pads of the target photovoltaic module.

[0167] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A monitoring and operation and maintenance method for a photovoltaic power station, characterized in that, Includes the following steps: S1. Perform perspective correction on the registration image of the grounding part to obtain the perspective-corrected image of the grounding part; extract the image area including the grounding pad, the bite position and the bite periphery from the perspective-corrected image of the grounding part to obtain the grounding part correction image. S2. Based on the grounding location correction image, obtain the rolled-up connection edge; S3. Accumulate the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, obtain the pre-rolling occupied area; S4. Based on the area occupied before the rollover, the current occlusal tooth area is extracted by regional overlap to obtain the cross-old trace occlusal bearing area. S5. Extract the area closest to the current biting tooth area to obtain the target bearing area; based on the cross-old mark biting bearing area and the target bearing area, monitor and maintain the grounding pad of the target photovoltaic module.

2. The monitoring and operation and maintenance method for a photovoltaic power station according to claim 1, characterized in that, Obtain a perspective-corrected image of the grounding area, including: Multi-directional image acquisition is performed on the location of the grounding pad in the target photovoltaic module to obtain multi-view images of the grounding part; Extract the positions of the grounding pad edge and component frame from multi-view images of the grounding part under different shooting directions; Based on the positions of the grounding pad edge and component frame under different shooting directions, image registration is performed on multi-view images of the grounding part to obtain the grounding part registered image; Extract the component border surface orientation from the registered image at the grounding location; Based on the surface orientation of the component frame, the registered image of the grounding part is subjected to perspective correction to obtain the perspective-corrected image of the grounding part.

3. The monitoring and operation and maintenance method for a photovoltaic power station according to claim 2, characterized in that, Obtain the grounding correction image, including: Based on the corresponding positions between the multi-view images of the grounding part, obtain the spatial depth information of each corresponding position in the perspective correction image of the grounding part; Determine the engagement position of the grounding pad; Based on the biting position, extract the area surrounding the biting position from the component border to obtain the biting perimeter area; The grounding correction image is obtained by cropping an image region including the grounding pad, the bite position, and the bite periphery from the perspective correction image of the grounding part, and writing the spatial depth information into the corresponding position in the image region.

4. The monitoring and operation and maintenance method for a photovoltaic power station according to claim 3, characterized in that, The rolled-up connection edges are obtained, including: Extract the concave contour formed by the pressing of the occlusal teeth of the grounding pad from the peri-occlusal area; Based on the concave contour, the old bite marks are obtained; The old bite marks were extracted in sheet-like areas to obtain the old bite mark rolled aluminum sheet; The edge where the old toothed rolled aluminum sheet meets the component frame is extracted to obtain the rolled connection edge.

5. A monitoring and maintenance method for a photovoltaic power station according to claim 4, characterized in that, The material extension length of each rolled material strip is obtained, including: Based on the rolled-up connection edge, the edge of the old toothed rolled-up aluminum sheet is extracted relative to the edge to obtain the rolled-up outer edge; Based on the connecting edge and the outer edge of the roll, the old toothed rolled aluminum sheet is divided into strips to obtain the rolled material strips; Based on the arrangement position of the rolled material strips on the rolled connection edge, the rolled material strips are arranged sequentially to obtain a rolled material strip sequence. Along the extension path from the connecting edge to the outer edge of each rolled material strip in the sequence of rolled material strips, obtain the spatial coordinates of the corresponding position of the rolled material strip; Based on the spatial coordinates of the corresponding positions of the rolled material strips, the spatial distances between adjacent positions in each rolled material strip are accumulated to obtain the material extension length of each rolled material strip.

6. The monitoring and maintenance method for a photovoltaic power station according to claim 5, characterized in that, The pre-rollout area is obtained, including: The position of each rolled material strip on the rolled connection edge is taken as the starting point for the spread of the rolled material strip; Based on the starting point and the length of the material extension, each rolled material strip in the rolled material strip sequence is rolled in reverse to obtain the restored area; Following the arrangement order of the rolled-up material strips, the restoration area is sequentially spliced ​​together to obtain the pre-rolled-up occupant area.

7. A monitoring and maintenance method for a photovoltaic power station according to claim 6, characterized in that, The area of ​​interlocking joint across old marks is obtained, including: Connect the current position of each rolled material strip in the rolled material strip sequence with the corresponding restoration area to obtain the material return path; Based on the material return path, the current position of each rolled material strip is mapped to the corresponding restoration area to obtain the old tooth mark backfill area; Get the current occlusal tooth region; Based on the current occlusal tooth area, the old tooth mark backfill area, and the pre-rollover occupied area, the current occlusal tooth area is extracted by regional overlap to obtain the occlusal bearing area across the old tooth mark.

8. A monitoring and maintenance method for a photovoltaic power station according to claim 7, characterized in that, The target receiving area is obtained, including: Remove the old bite marks and the pre-rollover area from the component border position to obtain the remaining border area; Using the outer contour of the current biting tooth area, perform morphological erosion on the remaining border area to obtain the area that the biting tooth can accommodate; Extract the region closest to the current occlusal tooth region from the region that can accommodate the occlusal teeth to obtain the target receiving region.

9. A monitoring and maintenance method for a photovoltaic power station according to claim 1, characterized in that, Monitoring and maintenance of the grounding pads of the target photovoltaic modules, including: Based on the interlocking area across the old marks, determine the current interlocking position of the grounding pad in the target photovoltaic module; Based on the target area, determine the reinstallation location of the grounding pad in the target photovoltaic module; Determine the position adjustment relationship of the grounding pad based on the current engagement position and the reinstallation position; By matching the position adjustment relationship of the grounding pads with the target photovoltaic modules, the monitoring and operation and maintenance results of the photovoltaic power station can be obtained; Based on the monitoring and maintenance results, the grounding pads of the target photovoltaic modules are monitored and maintained.

10. A system for monitoring and maintenance of a photovoltaic power plant according to any one of claims 1-9, characterized in that, The system includes: The calibration module performs perspective correction on the registration image of the grounding part to obtain a perspective-corrected image of the grounding part; it then extracts an image region including the grounding pad, the bite position, and the surrounding area of ​​the bite from the perspective-corrected image of the grounding part to obtain a grounding part calibration image. The image correction module, based on the grounding location, obtains the image of the image being flipped and connected. The in-situ restoration module accumulates the spatial distance between adjacent positions in each rolled material strip to obtain the material extension length of each rolled material strip; based on the material extension length, the pre-rolling occupied area is obtained; The occlusal reconstruction module extracts the overlapping areas of the current occlusal tooth area based on the pre-rollover occupied area, and obtains the cross-old trace occlusal bearing area. The operation and maintenance positioning module extracts the area closest to the current biting tooth area to obtain the target bearing area; based on the cross-old mark biting bearing area and the target bearing area, it monitors and maintains the grounding pads of the target photovoltaic module.

Images