Automatic identification of badmark points and control method and system for jump plates

Abstract

The present application relates to the technical field of jump control, and discloses a kind of Badmark point automatic identification and jump control method and system, wherein the method comprises: parsing Gerber file, determining the coordinate range and panel number of each panel;Identify the first panel in each panel Bad board identification template image and calculate the position offset vector;Calculate the array data of the remaining panels except the first panel;Calculate the RGB threshold range of the Badmark region under the good board state and establish the bad board determination interval and the good board determination interval;When the template matching score falls into the bad board determination interval, the corresponding panel number is written into the skip list, the method can automatically calculate the Badmark detection frame coordinates of all remaining panels, without manual calibration by operating personnel one by one panel, ensuring the accuracy and reliability of jump control.

Description

Technical Field

[0001] This invention relates to the field of springboard control technology, and in particular to a method and system for automatic identification of badmark points and control of springboards. Background Technology

[0002] In the electronics manufacturing industry, solder paste printing inspection of multi-panel PCBs is a crucial step in ensuring product quality. With the trend towards miniaturization and high density in products, multi-panel structures such as 8-panel and 16-panel PCBs are increasingly widely used in production. In actual production, the solder paste printing process may experience batch defects in individual panels due to factors such as wear on the printing press squeegee or stencil blockage. These defective panels are usually identified by operators after printing through visual inspection or simple marking.

[0003] However, existing AOI and SPI inspection equipment have significant technical shortcomings when processing multi-panel PCBs with marked defective boards. During the programming phase, operators need to manually move the camera to the defective board marking position for each panel to take a screenshot and mark it. When there are many panels, this method of marking each one is not only time-consuming and inefficient, but also prone to omissions or positional errors. Existing equipment lacks an intelligent recognition mechanism for defective board markings and cannot automatically adjust the inspection strategy based on the defective board marking information from the previous process. Summary of the Invention

[0004] This invention provides a method and system for automatic identification of badmark points and control of jumpers in a jigsaw puzzle. This invention can automatically calculate the coordinates of the badmark detection box for all remaining jigsaw puzzles, eliminating the need for operators to manually calibrate each puzzle individually, thus ensuring the accuracy and reliability of jumper control.

[0005] In a first aspect, the present invention provides a method for automatic identification of badmark points and control of jumpers in a jigsaw puzzle, the method comprising:

[0006] Parse the Gerber file to determine the coordinate range and panel number of each panel;

[0007] Identify the template image of the defective piece for the first piece in each puzzle and calculate its position offset vector;

[0008] Based on the position offset vector and the coordinate range of each panel, the Badmark detection box coordinates are calculated for the remaining panels except the first panel to obtain the array data;

[0009] Calculate the RGB threshold range of the Badmark region under the good board condition, and establish the bad board judgment interval and the good board judgment interval based on the bad board identification template image and the RGB threshold range;

[0010] After locating the Mark point on the target PCB board, the template matching score is calculated based on the array data. When the template matching score falls into the defective board judgment range, the corresponding panel number is written into the skip list.

[0011] In conjunction with the first aspect, in the first implementation of the first aspect of the present invention, the step of parsing the Gerber file to determine the coordinate range and panel number of each panel includes:

[0012] The Gerber file is parsed and the weld layer is filtered to obtain the weld layer graphic;

[0013] The region boundaries of each panel in the welded layer pattern are selected and the coordinates of the upper left and lower right corners of each panel are extracted to obtain the coordinate range of each panel.

[0014] The coordinate ranges of each puzzle piece are assigned numbers to obtain the puzzle piece numbers.

[0015] In conjunction with the first aspect, in a second implementation of the first aspect of the present invention, the step of identifying the defective board identifier template image of the first piece in each puzzle and calculating the position offset vector includes:

[0016] Extract the top-left and bottom-right coordinates of the first piece of each puzzle from the coordinate range of each puzzle piece, and calculate the corresponding center coordinates of the puzzle piece based on the top-left and bottom-right coordinates of the first piece of the puzzle piece;

[0017] Control the camera to move above the defective board marking area of ​​the first panel, adjust the screenshot frame to cover the defective board marking area, and then perform image acquisition to obtain the defective board marking template image and the center coordinates of the screenshot.

[0018] Subtract the X value of the center coordinate of the screenshot from the X value of the center coordinate of the panel to obtain the X-direction offset; subtract the Y value of the center coordinate of the screenshot from the Y value of the center coordinate of the panel to obtain the Y-direction offset.

[0019] The position offset vector is obtained by combining the X-direction offset and the Y-direction offset.

[0020] In conjunction with the first aspect, in a third implementation of the first aspect of the present invention, the step of calculating the Badmark detection box coordinates for the remaining panels (excluding the first panel) based on the position offset vector and the coordinate range of each panel to obtain array data includes:

[0021] The coordinate range of the remaining puzzle pieces (excluding the first piece) is traversed and the center coordinates are calculated to obtain the center coordinates of each remaining puzzle piece.

[0022] The center coordinates of each remaining panel are added to the X value and the Y value of the position offset vector to obtain the center coordinates of the Badmark detection box of each remaining panel.

[0023] The bad board identifier template image and screenshot frame size are copied to the center coordinates of each bad mark detection frame to establish a mapping relationship between the panel number and the center coordinates of the bad mark detection frame, thus obtaining array data.

[0024] In conjunction with the first aspect, in the fourth implementation of the first aspect of the present invention, the step of copying the bad board identifier template image and the screenshot frame size to the center coordinates of each bad mark detection frame, establishing a mapping relationship between the panel number and the center coordinates of the bad mark detection frame, and obtaining array data includes:

[0025] The RGB feature vector of the defective board identifier template image, the length value of the screenshot frame, and the width value are used as the feature parameter group;

[0026] At the positions corresponding to the center coordinates of each Badmark detection box, a rectangular detection box is generated according to the length and width values ​​of the screenshot box, and the feature parameter group is configured to each rectangular detection box;

[0027] Store each panel number, along with the corresponding Badmark detection box center coordinates and screenshot box size, as a record item. Combine all record items to obtain array data.

[0028] In conjunction with the first aspect, in the fifth implementation of the first aspect of the present invention, the step of calculating the RGB threshold range of the Badmark region under the good board state, and establishing a bad board determination interval and a good board determination interval based on the bad board identifier template image and the RGB threshold range, includes:

[0029] In the Badmark area of ​​the good board state, a sampling box is drawn and the RGB component values ​​of the pixels in the sampling box are extracted and the channel statistics are performed to obtain the R channel pixel value set, G channel pixel value set and B channel pixel value set;

[0030] Percentile calculations are performed on the R-channel pixel value set, G-channel pixel value set, and B-channel pixel value set respectively to obtain the RGB threshold range;

[0031] Set the bad board identifier template image as a negative sample matching template and set the RGB threshold range as a positive sample color space;

[0032] Based on the negative sample matching template and the positive sample color space, establish the bad board determination interval and the good board determination interval.

[0033] In conjunction with the first aspect, in the sixth implementation of the first aspect of the present invention, the step of establishing a bad board determination interval and a good board determination interval based on the negative sample matching template and the positive sample color space includes:

[0034] For each pixel in the Badmark region image to be detected, extract the RGB component values, count the number of pixels that simultaneously satisfy the following conditions: R channel value falls within the R threshold range, G channel value falls within the G threshold range, and B channel value falls within the B threshold range. Divide the number of pixels by the total number of pixels in the detection box to obtain the proportion of color matching pixels.

[0035] The image of the Badmark region to be detected is cross-correlated with the negative sample matching template to obtain the standard matching score;

[0036] Set a matching error threshold and a color proportion threshold. When the standard matching score is less than the matching error threshold and the color matching pixel proportion is greater than the color proportion threshold, it is judged as a good board. Set the first score range corresponding to the standard matching score as the good board judgment interval.

[0037] When the standard matching score is greater than or equal to the matching allowable error threshold or the color matching pixel ratio is less than or equal to the color ratio threshold, it is determined to be a bad board, and the second score range corresponding to the standard matching score is set as the bad board determination interval.

[0038] In conjunction with the first aspect, in the seventh implementation of the first aspect of the present invention, after locating the Mark point on the target PCB board, calculating the template matching score based on the array data, and when the template matching score falls into the defective board determination interval, writing the corresponding panel number into the skip list, includes:

[0039] After performing Mark point image matching and coordinate offset calculation on the target PCB board, the camera is driven to move sequentially to the Badmark position of each panel according to the Badmark detection box coordinates of each panel in the array data and the coordinate offset to acquire images and obtain the actual Badmark image of each panel.

[0040] Normalized cross-correlation matching calculations are performed on the actual Badmark images of each panel and the bad panel identification template images to obtain template matching scores;

[0041] The template matching score is input into the bad board judgment interval and the good board judgment interval for interval judgment. When the template matching score falls into the bad board judgment interval, the corresponding puzzle number is added to the skip list.

[0042] In conjunction with the first aspect, in the eighth implementation of the first aspect of the present invention, the automatic identification of badmark points and jump board control method for the puzzle board further includes:

[0043] During the component detection stage, the components to be detected are traversed in FOV path order. The coordinate position of the components to be detected and the coordinate range of each panel are determined to obtain the panel number to which the component to be detected belongs.

[0044] The panel number to which the component to be detected belongs is used as the query keyword to search and match in the skip list to determine whether the panel number exists in the skip list;

[0045] When the panel number to which the element to be detected belongs exists in the skip list, the image acquisition and algorithm calculation of the element to be detected are skipped, and the element to be detected is marked as skipped and untested in the detection result.

[0046] Secondly, the present invention provides an automatic identification and jumper control system for badmark points in a jigsaw puzzle, the automatic identification and jumper control system for badmark points in a jigsaw puzzle includes:

[0047] The parsing module is used to parse the Gerber file and determine the coordinate range and panel number of each panel;

[0048] The identification module is used to identify the defective board identifier template image of the first piece in each puzzle and calculate the position offset vector;

[0049] The calculation module is used to calculate the Badmark detection box coordinates of the remaining panels (excluding the first panel) based on the position offset vector and the coordinate range of each panel, so as to obtain array data.

[0050] A module is established to calculate the RGB threshold range of the Badmark region under the good board state, and to establish the bad board judgment interval and the good board judgment interval based on the bad board identification template image and the RGB threshold range;

[0051] The writing module is used to calculate the template matching score based on the array data after locating the Mark point on the target PCB board. When the template matching score falls into the bad board judgment range, the corresponding panel number is written into the skip list.

[0052] The technical solution provided by this invention achieves automatic identification and skipping control of bad board markers in multi-panel PCBs by establishing an automatic bad mark point array generation technology based on position offset vectors and a dual judgment model based on RGB color space and template matching. This effectively solves the technical problems of low programming efficiency and wasted time due to repeated bad board detection in existing technologies. By cropping the image of the bad board marker area of ​​the first panel and calculating its position offset vector relative to the panel center, combined with the coordinate range of each panel obtained from Gerber file parsing, the coordinates of the bad mark detection box of all remaining panels can be automatically calculated. This eliminates the need for manual calibration of each panel, and the programming efficiency is proportional to the number of panels, completely changing the inefficient traditional manual calibration mode. Regarding recognition accuracy, a dual-judgment model integrating color and shape features was established. Positive sample color space was obtained by RGB color sampling and threshold statistics of the Badmark region in the good board state. This was combined with a bad board identifier template image as a negative sample matching template. Two independent discrimination indicators were calculated: the proportion of color-matching pixels and the template similarity matching score. Only when both indicators simultaneously met the bad board judgment condition was the panel marked as bad. This dual-verification mechanism reduced the false judgment rate caused by single-feature discrimination. In terms of skip-board control implementation, Badmark recognition of all panels was pre-executed before actual detection, and the bad board numbers were written into a skip list. During the component detection stage, only the skip list needed to be consulted to determine whether to skip detection, avoiding redundant operations such as image acquisition and algorithm calculation for components on marked bad boards. The reduction in detection time was directly related to the proportion of bad boards. The mapping relationship table between panel numbers and Badmark detection box coordinates established in this invention ensured the accuracy and reliability of skip-board control.

[0053] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of an embodiment of the automatic identification of badmark points and jump board control method for the puzzle in this invention;

[0056] Figure 2 This is a schematic diagram of an embodiment of the automatic identification and jump board control system for badmark points in the puzzle, as described in this invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] The terms "comprising" and "having," and any variations thereof, used in the embodiments of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0059] To facilitate understanding of this embodiment, a detailed description of the automatic identification of badmark points and jump board control method for tile assembly disclosed in this embodiment of the invention will be provided first. For example... Figure 1 As shown, this method includes the following steps:

[0060] 101. Parse the Gerber file to determine the coordinate range and panel number of each panel;

[0061] Specifically, after the operator imports the Gerber file from the specified path using the offline programming software, the built-in Gerber syntax parsing module is invoked to perform graphical semantic reconstruction of the layer data contained in the Gerber file. This automatically identifies various layers and marks their attributes, such as solder layers, silkscreen layers, and drill layers. The solder layer graphic is the core processing object for panel division. The parsing result display interface lists all layer names and contents for the user to select. The operator selects the layer corresponding to the actual PCB solder surface and sets it as the "solder layer." The system automatically extracts the solder layer graphic data and presents it in the editing area, forming an interactive visual vector graphic. Using the software's graphic editing tools, the operator uses the mouse selection function to sequentially mark the rectangular area of ​​each panel along its outer edge. The system automatically records the coordinates of the upper left and lower right corners corresponding to each selection operation, forming the minimum and maximum boundaries of the panel area, thus clarifying the coordinate range of the current panel in the entire Gerber diagram. For panel layouts with a regular array structure, the operator activates the "automatic division" function. Based on the coordinates and position of the first panel, the theoretical permutation matrix is ​​calculated, and the actual position boundaries of each panel are identified using an image similarity analysis algorithm, quickly completing the boundary extraction operation for multiple panels. After all panel areas are extracted, each panel is assigned a unique number according to the selection or recognition order. The operator clicks the "OK" button to confirm the panel division, and the system automatically packages the panel coordinate range and number information into the program configuration file and saves it.

[0062] 102. Identify the defective board identifier template image of the first piece in each puzzle and calculate its position offset vector;

[0063] Specifically, the coordinates of the top-left and bottom-right corners of the first panel in each panel are extracted from the coordinate range of each panel. Averaging these two coordinates along the X and Y axes yields the geometric center of the current panel in the image coordinate system, which serves as the reference origin for offset calculation. Clicking on any pad of the first panel in the Gerber image activates that panel as the target area, initiating the motion control mechanism. The inspection platform, equipped with an industrial camera, is gradually moved to the area above the preset Badmark area on the current panel using the directional buttons. After confirming that the crosshair is accurately aligned with the center of the mark area, an adjustable screenshot tool is used to cover the entire Badmark area, with the frame boundary flush against the edge of the mark pattern, thus defining the spatial range of image acquisition. The coordinates of the geometric center point of the screenshot frame are recorded as the center reference position of the Badmark template image. Simultaneously, a color image within the screenshot area is acquired using the industrial camera and saved as a standardized Badmark template image file. The difference between the screenshot center coordinates and the panel's geometric center coordinates is calculated. Subtracting the X-coordinate of the panel center from the X-coordinate of the screenshot center yields the relative offset of the Badmark in the X direction; subtracting the Y-coordinate of the panel center from the Y-coordinate of the screenshot center yields the relative offset in the Y direction. The two offsets together constitute a two-dimensional position offset vector describing the spatial position difference of the Badmark image in the local coordinate system of the panel.

[0064] 103. Based on the position offset vector and the coordinate range of each panel, calculate the Badmark detection box coordinates for the remaining panels except the first panel to obtain the array data;

[0065] Specifically, the process iterates through all panel data in the panel number list except for the first one. For each panel, the X and Y values ​​from its top-left and bottom-right corners are averaged to obtain the geometric center coordinates of the current panel. The center position represents the theoretical geometric centroid of the panel in image space. The Badmark position offset vectors ΔX and ΔY are added to the X and Y coordinates of the current panel's center, respectively, to obtain the center coordinates of the Badmark detection box for each panel in the global coordinate system. These coordinates are the positioning anchor points of the Badmark detection box for the current panel in the image. The bad panel template image captured by the screenshot frame in the first panel, along with the corresponding rectangular area size parameters, are completely copied into the current detection box definition, forming a complete detection box description unit containing spatial coordinates, image template, and size specifications. Using the panel number as an index, a mapping relationship is established between the center coordinates of each Badmark detection box and its number, and these are stored uniformly in the array data structure. The array data structure contains the center position of the detection box of each panel, and also encapsulates algorithm configuration parameters such as template image path, color recognition threshold, and matching tolerance, forming a Badmark detection box array with fast loading and efficient recognition capabilities.

[0066] 104. Calculate the RGB threshold range of the Badmark area under the good board condition, and establish the bad board judgment interval and good board judgment interval based on the bad board identification template image and the RGB threshold range;

[0067] Specifically, color statistical modeling is performed on the good board areas that have not yet been labeled with bad board markers to construct the color recognition boundary for the normal board state. On any board confirmed to be without a bad board marker, the operator selects its corresponding Badmark detection box position and manually draws one or more rectangular sampling boxes within the image area covered by the Badmark detection box, ensuring the sampling area covers as many representative good board surface pixels as possible. After the sampling boxes are drawn, the RGB component values ​​of all pixels within each sampling box are extracted, and all R-channel, G-channel, and B-channel values ​​are constructed into independent value sets. Percentile statistical analysis is performed on the pixel set for each channel, using the 5th percentile as the lower threshold and the 95th percentile as the upper threshold, forming a positive sample enclosed cube in a three-dimensional color space. This color space represents the possible color fluctuation range of the normal good board area. Simultaneously, the bad board marker template image is used as the unique negative sample matching template. At the recognition logic level, two dimensions of judgment criteria are combined: if the RGB pixels of the Badmark region in the detected image are distributed outside the color space, it indicates that it does not meet the good board standard; if the normalized cross-correlation matching score between the detected image and the negative sample template is higher than a set threshold, it is judged to be similar to the bad board image. Image similarity matching score and color matching ratio are used as dual indicators to construct bad board and good board judgment intervals. The bad board judgment interval can be defined as a matching score in the high similarity segment and a color matching ratio below the threshold, while the good board judgment interval is defined as a high color matching ratio and a template matching score in the low similarity segment. Through joint modeling based on the negative sample template and the positive sample color domain, a highly discriminative image recognition and classification framework is constructed.

[0068] 105. After locating the Mark point on the target PCB board, calculate the template matching score based on the array data. When the template matching score falls into the defective board judgment range, write the corresponding panel number into the skip list.

[0069] Specifically, based on the Mark point recognition module, the detection platform moves the industrial camera to the preset Mark point position for image acquisition. Image matching is then performed with pre-stored Mark point image templates to calculate the global coordinate offset of the current actual PCB board relative to the programming stage. This global coordinate offset includes translational displacement and rotational angle errors. Based on this global coordinate offset, the center coordinates of the Badmark detection frame for each panel in the array data are corrected. This correction drives the XY-axis motion platform to move the camera one by one to the center position of the Badmark detection area for each panel. Upon reaching the target position, the camera completes real-time image acquisition under the current lighting configuration, and the acquired color image is used as the actual image of the current Badmark area of ​​the panel. A normalized cross-correlation image matching algorithm is used to perform similarity calculations between the acquired image and the Bad Board identifier template image, outputting a template matching score. A higher matching score indicates a greater similarity between the current area and the Bad Board identifier image. The template matching score is input into the recognition and judgment mechanism and compared with the bad board judgment interval and the good board judgment interval. If the template matching score falls into the bad board judgment interval, it means that the current puzzle has a highly suspected bad mark pattern, and the current puzzle number is written into the skip list.

[0070] In one specific embodiment, the process of performing step 101 may specifically include the following steps:

[0071] The Gerber file is parsed and the weld layer is filtered to obtain the weld layer graphic;

[0072] The region boundaries of each panel in the weld layer graphic are selected and the coordinates of the upper left and lower right corners of each panel are extracted to obtain the coordinate range of each panel;

[0073] The coordinate ranges of each puzzle piece are assigned numbers to obtain the puzzle piece numbers.

[0074] Specifically, the process involves importing and parsing Gerber format data using an offline programming software platform. The built-in Gerber parsing engine initiates a syntax analysis process, parsing and reconstructing the imported multi-layered design files according to format standards, extracting various layers and identifying their functional attributes. Layer information is displayed as a layer list on the interface. Operators identify the main layers corresponding to the physical welding surfaces, such as line layers named "TopLayer" or "BottomLayer," and set them as the base graphic for subsequent operations by clicking the "Set as Welding Layer" button. After confirming the welding layer, the vector graphic data carried by the welding layer is converted into a visual editing diagram, and the actual layout of all graphic elements is drawn with high precision in the Gerber graphics window. Operators manually define the effective area boundaries of each panel using the selection tool in the editing interface, i.e., by dragging the mouse to draw a rectangular bounding box, enclosing the graphic of each panel individually. Upon completion of each selection operation, the coordinates of the top-left and bottom-right corners of the selected area are automatically recorded. These are considered the minimum and maximum coordinate ranges of the current puzzle piece, respectively, establishing a two-dimensional coordinate description of the puzzle piece in the overall graphic space. In scenarios with a large number of puzzle pieces and regular arrangement, the operator can enable the "automatic division" function. Based on the first manually defined puzzle piece area, its size and center position are calculated. The theoretical positions of the remaining puzzle pieces are automatically deduced through horizontal and vertical stepping patterns. Subsequently, image correlation analysis algorithms are used to perform graphic matching verification on each deduced area, automatically confirming the effective boundaries of the puzzle pieces and extracting the corresponding coordinate points to form a set of puzzle piece coordinate ranges. After all puzzle piece areas are defined, a unique number is assigned to each puzzle piece according to the recognition order or image arrangement order. The number is generated in a row-column priority or absolute order incrementing manner, and each puzzle piece has a unique indexable number in subsequent recognition processes. In the internal data structure, the puzzle piece number is mapped one-to-one with its corresponding top-left and bottom-right corner coordinates to generate a puzzle piece coordinate index table. The puzzle piece order is then visually presented in the Gerber graph display interface in the form of numbered labels. After numbering, write the panel structure information into the program file and save it.

[0075] In one specific embodiment, the process of performing step 102 may specifically include the following steps:

[0076] Extract the top-left and bottom-right coordinates of the first piece in each puzzle from the coordinate range of each puzzle piece, and calculate the center coordinates of the corresponding puzzle piece based on the top-left and bottom-right coordinates of the first piece in each puzzle piece;

[0077] Move the camera above the defective board marking area of ​​the first panel, adjust the screenshot frame to cover the defective board marking area, and then capture the image to obtain the defective board marking template image and the center coordinates of the screenshot.

[0078] Subtract the X value of the center coordinate of the screenshot from the X value of the center coordinate of the panel to get the X-direction offset; subtract the Y value of the center coordinate of the screenshot from the Y value of the center coordinate of the panel to get the Y-direction offset.

[0079] By combining the offsets in the X and Y directions, the position offset vector is obtained.

[0080] Specifically, by accessing the number index list in the panel data structure, the panel with the smallest number or the first position is located, and its upper left and lower right corner coordinates are extracted from the coordinate range record of the first panel. Position calculations are performed on these two coordinate points in the horizontal and vertical dimensions to determine the center position of the first panel in the global graphic coordinate system. The operator switches to the jump area setting interface, manually selects the first panel area to activate its coordinate reference, and uses the F11 shortcut key or the arrow buttons to control the XY axis stepping drive system of the motion platform, moving the detection module equipped with a high-resolution industrial camera above the preset defective panel marking area. During the movement, the operator refers to the position of the crosshair in the camera's field of view image in real time, aligning its center with the geometric center of the defective panel marking area. After positioning is complete, the operator double-clicks the FOV area to activate the screenshot function, and manually scales and adjusts the screenshot area by dragging the frame boundary to ensure the rectangle completely covers the marking area without any extra background interference. Then, a second click on the screenshot frame completes the static image acquisition. The system automatically extracts the image region corresponding to the screenshot frame in the current camera view as a template image for identifying defective panels, and simultaneously records the coordinates of the center point of the screenshot frame in the image coordinate space as the reference point for the template center. After image acquisition, position offset calculation is performed. By comparing the horizontal and vertical differences between the screenshot center coordinates and the geometric center coordinates of the first panel, the relative displacement of the label position in the X and Y axes is obtained, forming two-dimensional offset parameters. The displacement results in the X and Y directions are combined and stored to generate a two-dimensional position offset vector.

[0081] In one specific embodiment, the process of performing step 103 may specifically include the following steps:

[0082] The coordinate range of the remaining puzzle pieces (excluding the first piece) is traversed and the center coordinates are calculated to obtain the center coordinates of each remaining puzzle piece.

[0083] The center coordinates of each remaining panel are summed with the X and Y values ​​of the position offset vector to obtain the center coordinates of the Badmark detection box for each remaining panel.

[0084] Copy the bad board identifier template image and screenshot frame size to the center coordinates of each bad mark detection frame, establish the mapping relationship between the panel number and the center coordinates of the bad mark detection frame, and obtain the array data.

[0085] Specifically, the first tile number that has completed template calibration is excluded from the internal tile data structure, and the set of numbers of the remaining tiles is obtained. Coordinate traversal is then performed on each tile in the set. During each traversal, the coordinates of the top-left and bottom-right corners of each tile are extracted from its data field. By calculating the median values ​​of these two boundary points in the horizontal and vertical dimensions, the center coordinates of the current tile in the global graphics space are obtained, which are the geometric reference points of the current tile in the image coordinate system. The two-dimensional position offset vector is added to the center horizontal and vertical coordinates of the current tile to perform the superposition operation of the relative displacement, deriving the center position of the Badmark detection box corresponding to the current tile. The center position of the Badmark detection box is the reference anchor point used for aligning the detection box during image recognition. After obtaining the center point of the Badmark detection box, the path of the bad tile identifier template image collected in the first tile and its width and height dimensions are copied, and these parameters, along with the center coordinates of the current detection box, are encapsulated into a Badmark detection box configuration unit. A unique ID is assigned to each Badmark detection box, corresponding one-to-one with the tile ID. This ID is stored as key-value pairs in the array data table, establishing a mapping between tile IDs and the center coordinates of the Badmark detection boxes. This mapping includes spatial coordinate information, as well as multi-dimensional recognition configuration fields such as image path, color judgment parameters, image matching threshold, and screenshot size. After generating all tile detection boxes, the array data is stored in a structured format in the program configuration file, and the center position of each tile's detection box is highlighted in the Gerber graph display interface.

[0086] In one specific embodiment, the process of copying the bad board identifier template image and screenshot frame size to the center coordinates of each bad mark detection frame, establishing a mapping relationship between the panel number and the center coordinates of the bad mark detection frame, and obtaining the array data can specifically include the following steps:

[0087] Use the RGB feature vector of the bad board identifier template image, the length and width of the screenshot frame as the feature parameter group;

[0088] At the location corresponding to the center coordinates of each Badmark detection box, a rectangular detection box is generated according to the length and width values ​​of the screenshot box, and the feature parameter group is configured to each rectangular detection box.

[0089] Store each panel number, along with the corresponding Badmark detection box center coordinates and screenshot box size, as a record item. Combine all record items to obtain array data.

[0090] Specifically, pixel-level analysis is performed on the bad mark template image to extract its RGB three-channel color distribution features, constructing an RGB feature vector including parameters such as color mean, channel variance, maximum value, and minimum value. Simultaneously, the length and width values ​​used by the screenshot frame during the template acquisition stage are recorded, forming size parameters representing the geometric shape of the rectangular detection frame. The RGB feature vector is combined with the length and width values ​​of the screenshot frame to form a feature parameter set. The center coordinates of each bad mark detection frame are read sequentially, and a rectangular area of ​​the same size as the template screenshot frame is constructed centered on the bad mark detection frame coordinates, serving as the actual analysis window during image recognition. This rectangular area is the detection frame, and its geometric boundary is jointly determined by the center coordinates and length and width parameters. The feature parameter set is bound to each detection frame. For each generated detection frame, its corresponding panel number, center coordinate position, and the length and width values ​​of the screenshot frame are recorded synchronously, and this information is structured and stored as a complete record item. All panel records are aggregated to form an array data containing all panel bad mark detection areas.

[0091] In one specific embodiment, the process of performing step 104 may specifically include the following steps:

[0092] In the Badmark area of ​​the good board state, a sampling box is drawn and the RGB component values ​​of the pixels in the sampling box are extracted and the channel statistics are performed to obtain the R channel pixel value set, G channel pixel value set and B channel pixel value set;

[0093] The percentiles of the R-channel pixel value set, G-channel pixel value set, and B-channel pixel value set are calculated to obtain the RGB threshold range;

[0094] Set the bad board identifier template image as the negative sample matching template and set the RGB threshold range to the positive sample color space;

[0095] The bad board determination interval and the good board determination interval are established based on the negative sample matching template and the positive sample color space.

[0096] Specifically, in the recognition interface, select any known good board area and manually draw one or more rectangular sampling boxes at the Badmark detection box location of that board using the mouse. These sampling boxes should cover the color features of representative good board areas as much as possible. After the sampling boxes are drawn, perform pixel-level extraction on the pixels within each sampling box, extracting the red, green, and blue components of each pixel separately, and classifying them by channel to form three independent sets of pixel values: the R channel pixel value set, the G channel pixel value set, and the B channel pixel value set. Perform statistical analysis on these three channel pixel sets, using percentile calculation to define the intervals for each set of values. By default, the 5th percentile is used as the lower threshold of the channel, and the 95th percentile is used as the upper threshold of the channel, obtaining the stable fluctuation range of the RGB channels in the actual application environment under the good board condition. The upper and lower limits of these three channels together constitute a cubic region in the RGB color space. The color space is regarded as the distribution boundary of good board samples in the color dimension and defined as the positive sample color space. Simultaneously, the defective board identification image collected in the first panel is used as a negative sample matching template. This template contains RGB feature vectors and typical defective board texture and structural information. Combining color recognition and image matching, a dual-channel judgment criterion is introduced into the logical judgment mechanism: if most pixels of the target detection image fall within the RGB positive sample color space and the template matching degree is low, it can be judged as a good board; if the target detection image deviates from the good board color space in color distribution and the template matching degree is highly similar to the defective board template, it is judged as a defective board. Based on this, defective board judgment intervals and good board judgment intervals are established. The defective board judgment interval corresponds to combinations with high matching degree but mismatched colors, while the good board judgment interval corresponds to combinations with good color matching but low template matching degree.

[0097] In one specific embodiment, the process of establishing the bad board determination interval and the good board determination interval based on the negative sample matching template and the positive sample color space can specifically include the following steps:

[0098] For each pixel in the Badmark region image to be detected, extract the RGB component values, count the number of pixels that simultaneously satisfy the following conditions: R channel value falls within the R threshold range, G channel value falls within the G threshold range, and B channel value falls within the B threshold range. Divide the number of pixels by the total number of pixels in the detection box to obtain the proportion of color matching pixels.

[0099] The image of the Badmark region to be detected is cross-correlated with the negative sample matching template to obtain the standard matching score;

[0100] Set the matching error threshold and the color proportion threshold. When the standard matching score is less than the matching error threshold and the color matching pixel proportion is greater than the color proportion threshold, it is judged as a good board. Set the first score range corresponding to the standard matching score as the good board judgment interval.

[0101] When the standard matching score is greater than or equal to the matching error threshold or the color matching pixel ratio is less than or equal to the color ratio threshold, it is judged as a bad board, and the second score range corresponding to the standard matching score is set as the bad board judgment interval.

[0102] Specifically, at each detection box location, each pixel in the region image is processed point-by-point, extracting the values ​​of its red, green, and blue color channels. All pixels are then filtered to count the number of pixels that simultaneously meet the following criteria: red channel value falls within the R threshold range, green channel value falls within the G threshold range, and blue channel value falls within the B threshold range. This number is then proportionally converted to the total number of pixels within the entire detection box area to obtain the percentage of color-matching pixels in the current image that match the good board color model in the color space, reflecting the overall consistency of the color in the current region. A normalized cross-correlation algorithm is used to perform image similarity analysis between the current detection image and a preset bad board identifier template image. A standard matching score is calculated using a sliding window and pixel normalization processing. The standard matching score quantifies the similarity between the test image and the bad board template; a lower score indicates a greater image difference. To achieve automatic judgment, a dual threshold mechanism is introduced, including a matching error threshold and a color proportion threshold, used to control the minimum standards for image matching strength and color consistency, respectively. When the standard matching score is less than the allowable matching error threshold and the proportion of color-matching pixels is greater than the color proportion threshold, the detected image is judged to be neither close to the bad board image template nor to have high color consistency with the good board. Therefore, it is judged as a good board, and the matching score range corresponding to this type of image is assigned to the first score range, i.e., the good board judgment interval. Conversely, when the standard matching score is greater than or equal to the allowable matching error threshold or the proportion of color-matching pixels is less than or equal to the color proportion threshold, the detected image is considered to have at least one of the two dimensions of shape or color that deviates significantly from the good board model and cannot meet the good board judgment criteria. Therefore, it is classified as a bad board, and its corresponding matching score range is assigned to the second score range, i.e., the bad board judgment interval.

[0103] The process involves performing normalized cross-correlation calculations between the image of the Badmark region to be detected and the negative sample matching template to obtain a standard matching score. This includes: representing the quantized pixel observations of the Badmark region image as the product of the true pixel value and the quantization coefficient; decomposing the nonlinear bias caused by the quantization effect into two independent terms: unknown matching parameters and unknown quantization coefficients; constructing a weighted least squares objective function based on the quantized pixel observations and the pixel values ​​of the negative sample matching template; performing a weighted summation of the squared pixel differences between the two images and setting a convergence criterion; performing a two-step recursive estimation of the unknown quantization coefficients and unknown matching parameters: first, estimating the quantization coefficients while fixing the matching parameters; second, estimating the matching parameters while fixing the quantization coefficients; iterating until the convergence criterion is met to obtain the optimal estimate; and finally, substituting the optimal estimate into the normalized cross-correlation formula to calculate the weighted correlation, resulting in the standard matching score after quantization error compensation.

[0104] The process includes establishing the bad board and good board judgment intervals, followed by an adaptive optimization process for threshold parameters based on fuzzy rules: Badmark recognition tests are performed on multiple sample PCBs, and the recognition accuracy and false positive rate are statistically analyzed under different combinations of matching allowable error thresholds and color proportion thresholds to obtain a parameter performance dataset; a fuzzy rule base is established, using color matching pixel proportion and standard matching score as fuzzy input variables and recognition confidence as a fuzzy output variable, describing the membership relationship between input and output using IF-THEN rules; iterative optimization calculations are performed on the parameter performance dataset, with the optimization objectives of maximizing recognition accuracy and minimizing false positive rate, adjusting the values ​​of the matching allowable error threshold and color proportion threshold; when the recognition accuracy reaches the preset accuracy target and the false positive rate is lower than the preset false positive rate upper limit, the current threshold combination is determined as the optimal parameter group and updated to the bad board and good board judgment intervals.

[0105] In one specific embodiment, the process of performing step 105 may specifically include the following steps:

[0106] After performing Mark point image matching and coordinate offset calculation on the target PCB board, the camera is driven to move sequentially to the Badmark position of each panel according to the Badmark detection box coordinates and coordinate offset of each panel in the array data to acquire images and obtain the actual Badmark image of each panel.

[0107] Normalized cross-correlation matching calculations are performed on the actual Badmark images of each panel and the bad panel identification template images to obtain the template matching score;

[0108] The template matching score is input into the bad board judgment interval and the good board judgment interval for interval judgment. When the template matching score falls into the bad board judgment interval, the corresponding piece number is added to the skip list.

[0109] Specifically, after the PCB is fed into the system, a baseline positioning operation is performed. A motion platform moves the camera to a preset Mark point position, acquires Mark point images, and performs image matching processing with standard Mark patterns stored in the program file. This yields the coordinate offset of the current target PCB board relative to the programming stage in its actual clamping state, including X-direction offset, Y-direction offset, and rotation angle changes. After calculating the overall offset, the center coordinates of the Badmark detection frame corresponding to each panel number in the array data structure are corrected. The offset is applied to the center coordinates of each detection frame, obtaining the actual sampling position of the Badmark area of ​​each panel in the global coordinate system during actual inspection. The motion control program is then started, driving the industrial camera platform to move to each Badmark detection point one by one according to the corrected detection frame coordinates. The light source is illuminated, and a color image of the current area is acquired. This image serves as the actual Badmark area image sample of the panel in the current inspection cycle. After image acquisition, the image recognition module performs normalized cross-correlation image matching calculations on each Badmark region image and a pre-defined bad board identifier template image. Through pixel-level alignment, grayscale normalization, and correlation analysis using local window sliding, a standardized template matching score is obtained. A higher template matching score indicates a greater similarity between the current image and the bad board template. The template matching score is then compared with the good board and bad board judgment intervals set during the modeling process. If the current matching score falls within the bad board judgment interval, the current panel is considered to have obvious Badmark identifier features, and its image features are highly similar to the pre-defined bad board template, failing to meet the good board judgment criteria. The panel's number is then added to the skip list.

[0110] In one specific embodiment, the automatic identification of badmark points and jump board control method for puzzle pieces further includes the following steps:

[0111] During the component inspection phase, the components to be inspected are traversed in FOV path order. The coordinate position of the components to be inspected and the coordinate range of each panel are determined to obtain the panel number to which the component to be inspected belongs.

[0112] The panel number to which the component to be tested belongs is used as the query keyword to search and match in the skip list to determine whether the panel number exists in the skip list;

[0113] When the panel number to which the element to be detected belongs exists in the skip list, the image acquisition and algorithm calculation of the element to be detected are skipped, and the element to be detected is marked as skipped and untested in the detection results.

[0114] Specifically, after completing Badmark recognition and skipping panel number recording, the component detection stage begins. Following a predetermined FOV path sequence, the image coordinate plan for the detection area is loaded frame by frame, and the motion platform moves the camera to acquire corresponding images sequentially. At each FOV path node, the coordinate data of all components to be detected within the current field of view is extracted, and spatial attribution is determined for each component. The X and Y coordinates of the component are compared with the coordinate range of all panels to determine if it falls within the rectangular bounding box defined by the upper left and lower right corner coordinates of a particular panel. If a panel area is matched, the panel number is marked as the panel number to which the current component belongs, and an attribution mapping relationship between the component and the panel number is established in the internal cache structure. After attribution determination, the skipping panel control module is called, using the component's panel number as the query keyword to match and search the skip list generated earlier by the Badmark recognition process to determine if the current panel number exists in the skip records. If a match is successful, it means that the panel was marked as an invalid area in an earlier stage due to the detection of a bad panel identifier, and therefore will not participate in any subsequent detection processes. Once it is confirmed that the panel number does exist in the skip list, the skip logic is executed for the current component's detection task. The skip logic includes not triggering the camera to acquire images, not calling any image processing algorithms, not executing the judgment process, not generating a detection score, and only setting the component's status field to "skipped and not tested" in the detection result data structure, while simultaneously recording its associated panel number and skip reason code.

[0115] The above describes the automatic identification of badmark points and the jump board control method in the embodiments of the present invention. The following describes the automatic identification of badmark points and the jump board control system in the embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the automatic identification and jump board control system for badmark points in the puzzle is as follows:

[0116] Parsing module 201 is used to parse the Gerber file and determine the coordinate range and panel number of each panel;

[0117] The identification module 202 is used to identify the defective board identifier template image of the first piece in each puzzle and calculate the position offset vector;

[0118] The calculation module 203 is used to calculate the Badmark detection box coordinates of the remaining panels except the first panel based on the position offset vector and the coordinate range of each panel, so as to obtain the array data.

[0119] Module 204 is established to calculate the RGB threshold range of the Badmark region under the good board state, and to establish the bad board judgment interval and the good board judgment interval based on the bad board identification template image and the RGB threshold range;

[0120] The writing module 205 is used to calculate the template matching score based on the array data after locating the Mark point on the target PCB board. When the template matching score falls into the bad board judgment range, the corresponding panel number is written to the skip list.

[0121] Through the collaborative efforts of the aforementioned components, and by establishing an automatic badmark point array generation technology based on position offset vectors and a dual judgment model based on RGB color space and template matching, automatic identification and skipping control of bad board markers in multi-panel PCBs are achieved. This effectively solves the technical problems of low programming efficiency and wasted time due to repeated bad board detection in existing technologies. By cropping the image of the bad board marker area of ​​the first panel and calculating its position offset vector relative to the panel center, combined with the coordinate range of each panel obtained from Gerber file parsing, the coordinates of the badmark detection box for all remaining panels can be automatically calculated. This eliminates the need for manual calibration of each panel, and the programming efficiency is proportionally improved with the number of panels, completely changing the inefficient traditional manual calibration mode. Regarding recognition accuracy, a dual-judgment model integrating color and shape features was established. Positive sample color space was obtained by RGB color sampling and threshold statistics of the Badmark region in the good board state. This was combined with a bad board identifier template image as a negative sample matching template. Two independent discrimination indicators were calculated: the proportion of color-matching pixels and the template similarity matching score. Only when both indicators simultaneously met the bad board judgment condition was the panel marked as bad. This dual-verification mechanism reduced the false judgment rate caused by single-feature discrimination. In terms of skip-board control implementation, Badmark recognition of all panels was pre-executed before actual detection, and the bad board numbers were written into a skip list. During the component detection stage, only the skip list needed to be consulted to determine whether to skip detection, avoiding redundant operations such as image acquisition and algorithm calculation for components on marked bad boards. The reduction in detection time was directly related to the proportion of bad boards. The mapping relationship table between panel numbers and Badmark detection box coordinates established in this invention ensured the accuracy and reliability of skip-board control.

[0122] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0123] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatic identification of badmark points and control of jump boards in a tile assembly, characterized in that, include: Parse the Gerber file to determine the coordinate range and panel number of each panel; Identify the template image of the defective piece for the first piece in each puzzle and calculate its position offset vector; Based on the position offset vector and the coordinate range of each panel, the Badmark detection box coordinates are calculated for the remaining panels except the first panel to obtain the array data; Calculate the RGB threshold range of the Badmark region under the good board state, and set the bad board identifier template image as the negative sample matching template and the RGB threshold range as the positive sample color space; extract the RGB component values ​​of each pixel in the Badmark region image to be detected, count the number of pixels that simultaneously satisfy the following conditions: R channel value falls within the R threshold range, G channel value falls within the G threshold range, and B channel value falls within the B threshold range, and divide the number of pixels by the total number of pixels in the detection box to obtain the proportion of color matching pixels; The image of the Badmark region to be detected is cross-correlated with the negative sample matching template using normalized methods to obtain a standard matching score. A matching error threshold and a color proportion threshold are set. When the standard matching score is less than the matching error threshold and the color matching pixel proportion is greater than the color proportion threshold, it is determined to be a good board, and a first score range corresponding to the standard matching score is set as the good board determination interval. When the standard matching score is greater than or equal to the matching error threshold or the color matching pixel proportion is less than or equal to the color proportion threshold, it is determined to be a bad board, and a second score range corresponding to the standard matching score is set as the bad board determination interval. After locating the Mark point on the target PCB board, the template matching score is calculated based on the array data. When the template matching score falls into the defective board judgment range, the corresponding panel number is written into the skip list.

2. The method for automatic identification of badmark points and control of skips in a mosaic as described in claim 1, characterized in that, The process of parsing the Gerber file to determine the coordinate range and panel number of each panel includes: The Gerber file is parsed and the weld layer is filtered to obtain the weld layer graphic; The region boundaries of each panel in the welded layer pattern are selected and the coordinates of the upper left and lower right corners of each panel are extracted to obtain the coordinate range of each panel. The coordinate ranges of each puzzle piece are assigned numbers to obtain the puzzle piece numbers.

3. The method for automatic identification of badmark points and control of skips in a mosaic as described in claim 1, characterized in that, The process of identifying the defective board identifier template image of the first board in each puzzle and calculating its position offset vector includes: Extract the top-left and bottom-right coordinates of the first piece of each puzzle from the coordinate range of each puzzle piece, and calculate the corresponding center coordinates of the puzzle piece based on the top-left and bottom-right coordinates of the first piece of the puzzle piece; Control the camera to move above the defective board marking area of ​​the first panel, adjust the screenshot frame to cover the defective board marking area, and then perform image acquisition to obtain the defective board marking template image and the center coordinates of the screenshot. Subtract the X value of the center coordinate of the screenshot from the X value of the center coordinate of the panel to obtain the X-direction offset; subtract the Y value of the center coordinate of the screenshot from the Y value of the center coordinate of the panel to obtain the Y-direction offset. The position offset vector is obtained by combining the X-direction offset and the Y-direction offset.

4. The method for automatic identification of badmark points and control of jump boards according to claim 1, characterized in that, The step involves calculating the Badmark detection box coordinates for the remaining panels (excluding the first panel) based on the position offset vector and the coordinate range of each panel, to obtain array data, including: The coordinate range of the remaining puzzle pieces (excluding the first piece) is traversed and the center coordinates are calculated to obtain the center coordinates of each remaining puzzle piece. The center coordinates of each remaining panel are added to the X value and the Y value of the position offset vector to obtain the center coordinates of the Badmark detection box of each remaining panel. The bad board identifier template image and screenshot frame size are copied to the center coordinates of each bad mark detection frame to establish a mapping relationship between the panel number and the center coordinates of the bad mark detection frame, thus obtaining array data.

5. The method for automatic identification of badmark points and control of skips in a mosaic as described in claim 4, characterized in that, The process involves copying the bad board identifier template image and screenshot frame size to the center coordinates of each bad mark detection frame, establishing a mapping relationship between the panel number and the center coordinates of the bad mark detection frame, and obtaining array data, including: The RGB feature vector of the defective board identifier template image, the length value of the screenshot frame, and the width value are used as the feature parameter group; At the positions corresponding to the center coordinates of each Badmark detection box, a rectangular detection box is generated according to the length and width values ​​of the screenshot box, and the feature parameter group is configured to each rectangular detection box; Store each panel number, along with the corresponding Badmark detection box center coordinates and screenshot box size, as a record item. Combine all record items to obtain array data.

6. The method for automatic identification of badmark points and control of skips in a mosaic as described in claim 1, characterized in that, The calculation of the RGB threshold range of the Badmark region under the good board condition includes: In the Badmark area of ​​the good board state, a sampling box is drawn and the RGB component values ​​of the pixels in the sampling box are extracted and the channel statistics are performed to obtain the R channel pixel value set, G channel pixel value set and B channel pixel value set; The percentiles of the R-channel pixel value set, G-channel pixel value set, and B-channel pixel value set are calculated to obtain the RGB threshold range.

7. The method for automatic identification of badmark points and control of jump boards according to claim 6, characterized in that, After locating the Mark point on the target PCB board, a template matching score is calculated based on the array data. When the template matching score falls within the defective board determination range, the corresponding panel number is written into the skip list, including: After performing Mark point image matching and coordinate offset calculation on the target PCB board, the camera is driven to move sequentially to the Badmark position of each panel according to the Badmark detection box coordinates of each panel in the array data and the coordinate offset to acquire images and obtain the actual Badmark image of each panel. Normalized cross-correlation matching calculations are performed on the actual Badmark images of each panel and the bad panel identification template images to obtain template matching scores; The template matching score is input into the bad board judgment interval and the good board judgment interval for interval judgment. When the template matching score falls into the bad board judgment interval, the corresponding puzzle number is added to the skip list.

8. The method for automatic identification of badmark points and control of skips in a mosaic as described in claim 1, characterized in that, The automatic identification of badmark points and jump board control method for the mosaic also includes: During the component detection stage, the components to be detected are traversed in FOV path order. The coordinate position of the components to be detected and the coordinate range of each panel are determined to obtain the panel number to which the component to be detected belongs. The panel number to which the component to be detected belongs is used as the query keyword to search and match in the skip list to determine whether the panel number exists in the skip list; When the panel number to which the element to be detected belongs exists in the skip list, the image acquisition and algorithm calculation of the element to be detected are skipped, and the element to be detected is marked as skipped and untested in the detection result.

9. A panel badmark point automatic identification and jump board control system, characterized in that, The method for automatically identifying badmark points and controlling jumpers as described in any one of claims 1-8 includes: The parsing module is used to parse the Gerber file and determine the coordinate range and panel number of each panel; The identification module is used to identify the defective board identifier template image of the first piece in each puzzle and calculate the position offset vector; The calculation module is used to calculate the Badmark detection box coordinates of the remaining panels (excluding the first panel) based on the position offset vector and the coordinate range of each panel, so as to obtain array data. A module is established to calculate the RGB threshold range of the Badmark region under the good board state, and to establish the bad board judgment interval and the good board judgment interval based on the bad board identification template image and the RGB threshold range; The writing module is used to calculate the template matching score based on the array data after locating the Mark point on the target PCB board. When the template matching score falls into the bad board judgment range, the corresponding panel number is written into the skip list.

Images