A high-precision cutting and positioning method and system for patch back glue based on machine vision

Abstract

The present invention relates to the field of machine vision technology, and discloses a high-precision cutting and positioning method and system for patch back glue based on machine vision, including: collecting the surface image of the patch back glue, performing global edge division and local center division on the surface image of the patch back glue, performing a detail scaling operation on the surface image of the patch back glue after global edge division and local center division, fusing the global back glue surface image and the local back glue surface image to obtain the patch back glue image; identifying the three-dimensional information of the patch back glue image, determining the cutting plane of the patch back glue image according to the three-dimensional information; identifying the bidirectional cutting direction corresponding to the cutting plane, generating a bidirectional cutting path of the patch back glue image according to the bidirectional cutting direction and the cutting plane; generating a cutting connection diagram of the patch back glue image according to the bidirectional cutting path, and positioning the cutting position of the patch back glue image through the cutting connection diagram. The present invention can improve the accuracy during the cutting and positioning of the patch back glue.

Description

Technical Field

[0001] The present invention relates to the technical field of machine vision, and particularly to a high-precision cutting and positioning method and system for patch back glue based on machine vision. Background Art

[0002] With the continuous development of electronic products towards miniaturization, thinness, lightness and high performance, the size of electronic components is constantly shrinking. As a key material connecting electronic components and circuit boards, the cutting accuracy of patch back glue directly affects the accuracy and stability of component installation. Only by achieving high-precision cutting and positioning of patch back glue can a large number of tiny components be accurately placed in a limited space to ensure the stable operation of equipment performance.

[0003] The existing cutting and positioning technology for patch back glue obtains the surface information of the back glue based on image acquisition technology. In practical applications, traditional image acquisition may have a perspective limitation and cannot comprehensively obtain the surface information of the back glue, resulting in inaccurate positioning. In addition, the processing of the glue image is not fine enough, and it is difficult to clearly distinguish the edges and details, affecting the positioning accuracy, so the accuracy of cutting and positioning patch back glue is relatively low. Summary of the Invention

[0004] The present invention provides a high-precision cutting and positioning method and system for patch back glue based on machine vision, and its main purpose is to solve the problem of relatively low accuracy in cutting and positioning patch back glue.

[0005] To achieve the above object, a high-precision cutting and positioning method for patch back glue based on machine vision provided by the present invention includes:

[0006] Collecting the surface image of the patch back glue at all-round angles based on a preset image acquisition, performing global edge division on the surface image of the patch back glue, and performing a global detail scaling operation on the surface image of the patch back glue after global edge division to obtain a global back glue surface image;

[0007] Performing local center division on the global back glue surface image, performing a local detail scaling operation on the global back glue surface image after local center division to obtain a local back glue surface image, and fusing the global back glue surface image and the local back glue surface image to obtain a patch back glue image;

[0008] Identifying the three-dimensional information of the patch back glue image by using a preset dual vision positioning algorithm, and determining the cutting plane of the patch back glue image according to the three-dimensional information;

[0009] Identifying the two-way cutting direction corresponding to the cutting plane, and generating a two-way cutting path of the patch back glue image according to the two-way cutting direction and the cutting plane;

[0010] Generate a cutting connection diagram of the patch back glue image according to the bidirectional cutting path, and locate the cutting position of the patch back glue image through the cutting connection diagram.

[0011] Optionally, the method for collecting the surface image of the patch back glue based on the preset all-round image collection angles includes:

[0012] Extract the horizontal direction and the vertical direction corresponding to the all-round image collection angles;

[0013] Divide the all-round image collection angles into horizontal all-round angles according to the preset horizontal angle intervals in the horizontal direction;

[0014] Divide the all-round image collection angles into vertical all-round angles according to the preset vertical angle intervals in the vertical direction;

[0015] Configure all-round image collection points according to the horizontal all-round angles and the vertical all-round angles, and collect the surface image of the patch back glue by using the all-round image collection points.

[0016] Optionally, the method for globally partitioning the edge of the surface image of the patch back glue includes:

[0017] Convert the surface image of the patch back glue into a back glue surface grayscale image;

[0018] Identify the pixel points of the back glue surface grayscale image, and generate the edge intensity distribution of the back glue surface grayscale image according to the pixel points;

[0019] Identify edge points and non-edge points by using the edge intensity distribution and the preset edge threshold;

[0020] Determine the global edge of the surface image of the patch back glue according to the edge points and the non-edge points.

[0021] Optionally, the method for performing a global detail scaling operation on the surface image of the patch back glue after global edge partitioning to obtain a global back glue surface image includes:

[0022] Extract the global edge corresponding to the surface image of the patch back glue after global edge partitioning;

[0023] Identify the initial edge points of the global edge, and calculate the dynamic edge point coefficient of the global edge according to the initial edge points: Wherein, is the dynamic edge point coefficient, is the gray scale change weight, is the edge change weight, is the initial edge point is the number of pixel points in the neighborhood of, is the pixel point in the neighborhood The grayscale value of is the average grayscale of the pixel points in the neighborhood. is the sine function. is the arctangent function. is the target candidate edge point;

[0024] Determine the global edge points of the global edge according to the dynamic edge point coefficient;

[0025] Perform local magnification operations on the global edge points one by one, identify the abnormal images of the patch adhesive surface image according to the magnified global edge points, and perform multi-dimensional enhancement processing on the abnormal images to obtain the edge surface images corresponding to each global edge point;

[0026] Perform shrinkage processing on the edge surface images, identify the edge intensity distribution of the edge images after shrinkage processing, and perform multi-dimensional enhancement processing on the edge images according to the edge intensity distribution to obtain the global adhesive surface image.

[0027] Optionally, the local center division of the global adhesive surface image includes:

[0028] Determine the initial local division window according to the image shape of the global adhesive surface image;

[0029] Determine the division area line of the global adhesive surface image according to the initial local division window;

[0030] Identify the two-way dynamic window factor of the initial local division window through the division area line;

[0031] Dynamically update the initial local division window using the two-way dynamic window factor, and return to the step of determining the division area line of the global adhesive surface image according to the initial local division window until the global adhesive surface image is completely divided;

[0032] When the global adhesive surface image is completely divided, determine the local area of the global adhesive surface image according to each division area line.

[0033] Optionally, the fusion of the global adhesive surface image and the local adhesive surface image to obtain the patch adhesive image includes:

[0034] Identify the local edge features of the local adhesive surface image and the global edge features of the global adhesive surface image;

[0035] Compare the feature positions of the local edge features with the feature positions of the global edge features to obtain the same feature positions;

[0036] Fuse the local edge features and the global edge features according to the same feature positions to obtain the patch adhesive image.

[0037] Optionally, in the embodiments of the present invention, the identifying the three-dimensional information of the patch adhesive image by using a preset dual-vision positioning algorithm includes:

[0038] Identifying the shape complexity index and the feature stability index of the patch adhesive image;

[0039] Converting the world coordinates of the patch adhesive image into image plane coordinates;

[0040] Determining a first weight and a second weight according to the shape complexity index and the feature stability index, where the first weight and the second weight are: Wherein, is the first weight, is the second weight, is the shape complexity index, is the stability index of the machine vision feature, is the stability index of the laser sensor feature;

[0041] Using the following dual-vision positioning algorithm to perform three-dimensional fusion on the image plane coordinates and the pre-acquired image measurement height to obtain the three-dimensional information of the patch adhesive image: Wherein, is the abscissa information in the three-dimensional information, is the ordinate information in the three-dimensional information, is the vertical coordinate information in the three-dimensional information, is the first weight, is the second weight, is the plane abscissa of the patch adhesive target point in the world coordinate system, is the plane ordinate of the patch adhesive target point in the world coordinate system, is the height coordinate of the patch adhesive target point, is the projection abscissa of the laser measurement point on the plane, is the projection ordinate of the laser measurement point on the plane, is the target height corresponding to the laser measurement point.

[0042] Optionally, the generating the two-way cutting path of the patch adhesive image according to the two-way cutting direction and the cutting plane includes:

[0043] Randomly select any one of the cutting planes as the initial cutting plane, and gradually move from any endpoint in the initial cutting plane according to the two-way cutting direction to obtain an initial cutting path;

[0044] Extend the initial cutting path according to the first cutting direction in the bidirectional cutting direction and the cutting sequence of the cutting plane to obtain a first cutting path;

[0045] Extend the initial cutting path according to the second cutting direction in the bidirectional cutting direction and the cutting sequence to obtain a second cutting path;

[0046] Generate a bidirectional cutting path of the patch adhesive image according to the first cutting path and the second cutting path.

[0047] Optionally, generating a cutting connection diagram of the patch adhesive image according to the bidirectional cutting path includes:

[0048] Discretize the bidirectional cutting path to obtain path discrete points;

[0049] Identify the intersection points in the path discrete points, and determine the connection relationship between the path discrete points according to the positions corresponding to the intersection points;

[0050] Connect the path discrete points according to the connection relationship to obtain a cutting connection diagram.

[0051] To solve the above problems, the present invention also provides a high-precision cutting positioning system for patch adhesive based on machine vision. The system includes:

[0052] An image division module for collecting a surface image of the patch adhesive based on a preset all-round angle of image acquisition, performing global edge division on the surface image of the patch adhesive, and performing a global detail scaling operation on the surface image of the patch adhesive after global edge division to obtain a global surface image of the adhesive;

[0053] An image fusion module for performing local center division on the global surface image of the adhesive, performing a local detail scaling operation on the global surface image of the adhesive after local center division to obtain a local surface image of the adhesive, and fusing the global surface image of the adhesive and the local surface image of the adhesive to obtain a patch adhesive image;

[0054] A cutting plane determination module for identifying three-dimensional information of the patch adhesive image by using a preset double vision positioning algorithm, and determining a cutting plane of the patch adhesive image according to the three-dimensional information;

[0055] A bidirectional cutting path generation module for identifying a bidirectional cutting direction corresponding to the cutting plane, and generating a bidirectional cutting path of the patch adhesive image according to the bidirectional cutting direction and the cutting plane;

[0056] The cutting position positioning module is used to generate a cutting connection diagram of the patch back glue image according to the bidirectional cutting path, and position the cutting position of the patch back glue image through the cutting connection diagram.

[0057] In the embodiment of the present invention, by collecting images in all directions, the complete surface information of the patch back glue can be obtained. The global edge division and detail scaling make the overall features of the patch back glue clearer, which is beneficial to grasping the overall contour and edge details; the local center division and detail scaling focus on the key areas and highlight the subtle features; the fusion of the global and local back glue surface images combines the advantages of the whole and the part, and the generated patch back glue image information is more abundant and accurate; the dual-vision positioning algorithm recognizes three-dimensional information, can accurately determine the cutting plane, and significantly improves the cutting accuracy; recognizing the bidirectional cutting direction and generating a path optimizes the cutting process, making the cutting more efficient and reasonable; the generated cutting connection diagram can visually and clearly present the cutting position. Therefore, the high-precision cutting positioning method and system for patch back glue based on machine vision proposed by the present invention can solve the problem of low accuracy in cutting and positioning the patch back glue. Brief Description of the Drawings

[0058] Figure 1 It is a schematic flow chart of a high-precision cutting and positioning method for patch back glue based on machine vision provided by an embodiment of the present invention;

[0059] Figure 2 It is a functional module diagram of a high-precision cutting and positioning system for patch back glue based on machine vision provided by an embodiment of the present invention.

[0060] The realization, functional characteristics and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed Embodiments

[0061] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0062] The embodiments of the present application provide a high-precision cutting and positioning method for patch back glue based on machine vision. The execution subject of the high-precision cutting and positioning method for patch back glue based on machine vision includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiments of the present application. In other words, the high-precision cutting and positioning method for patch back glue based on machine vision can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0063] Referring to Figure 1 As shown, it is a schematic flowchart of a high-precision cutting and positioning method for patch back glue based on machine vision provided by an embodiment of the present invention. In this embodiment, the high-precision cutting and positioning method for patch back glue based on machine vision includes:

[0064] S1. Collect a surface image of the patch back glue based on a preset all-round image acquisition angle, perform global edge division on the surface image of the patch back glue, and perform a global detail scaling operation on the surface image of the patch back glue after global edge division to obtain a global back glue surface image.

[0065] In the embodiments of the present invention, the surface image of the patch back glue refers to a two-dimensional image obtained by photographing the surface of the patch back glue through an image acquisition device (such as a camera).

[0066] In the embodiments of the present invention, collecting a surface image of the patch back glue based on a preset all-round image acquisition angle includes:

[0067] Extract the horizontal direction and the vertical direction corresponding to the all-round image acquisition angle;

[0068] Divide the all-round image acquisition angle into horizontal all-round angles according to a preset horizontal angle interval in the horizontal direction;

[0069] Divide the all-round image acquisition angle into vertical all-round angles according to a preset vertical angle interval in the vertical direction;

[0070] Configure all-round image acquisition points according to the horizontal all-round angles and the vertical all-round angles, and use the all-round image acquisition points to collect a surface image of the patch back glue.

[0071] Specifically, when performing image acquisition, based on a pre-determined reference coordinate system, the horizontal direction and the vertical direction are selected as two basic dimensions. For example, the horizontal direction is defined as the transverse direction of the camera's shooting plane, and the vertical direction is defined as the longitudinal direction. The preset horizontal angle interval is a custom value used to determine the step size of the angle change when acquiring images in the horizontal direction. For example, the preset horizontal angle interval is 10 degrees. In the horizontal direction, starting from a starting angle (assumed to be 0 degrees), the angles are divided at intervals of 0 degrees in sequence to obtain the horizontal omnidirectional angles. Similar to the horizontal direction, in the vertical direction, it is also divided according to the preset vertical angle interval to obtain the vertical omnidirectional angles.

[0072] Specifically, the horizontal omnidirectional angles and the vertical omnidirectional angles together constitute the omnidirectional acquisition angle range, and the omnidirectional image acquisition points are determined by the combination of the horizontal omnidirectional angles and the vertical omnidirectional angles. For example, when the horizontal angle is 30 degrees and the vertical angle is 45 degrees, a specific acquisition point position is determined. By combining all the horizontal omnidirectional angles and the vertical omnidirectional angles, omnidirectional image acquisition points are obtained. Then, the camera or other image acquisition devices are placed at the positions of the acquisition points and photographed at the corresponding angles, so as to acquire images of the surface of the patch adhesive at different angles.

[0073] Furthermore, in order to improve the image quality of the image of the surface of the patch adhesive and provide a high-quality image basis for positioning, it is necessary to analyze the edge image of the image of the surface of the patch adhesive.

[0074] In the embodiment of the present invention, global edge division refers to processing the edges of the entire image of the surface of the patch adhesive, extracting the edge information in the image, and clearly and accurately representing the overall boundary of the patch adhesive.

[0075] In the embodiment of the present invention, the global edge division of the image of the surface of the patch adhesive includes:

[0076] Converting the image of the surface of the patch adhesive into a grayscale image of the adhesive surface;

[0077] Identifying the pixel points of the grayscale image of the adhesive surface and generating the edge intensity distribution of the grayscale image of the adhesive surface according to the pixel points;

[0078] Using the edge intensity distribution and a preset edge threshold to identify edge points and non-edge points;

[0079] Determining the global edge of the image of the surface of the patch adhesive according to the edge points and the non-edge points.

[0080] Specifically, the surface image of the patch adhesive is converted into a grayscale image, that is, the color information of each pixel point in the color image is represented by a grayscale value, which can simplify the image data while retaining useful information such as the brightness and contrast of the image for edge detection, that is, convert the RGB value into a grayscale value to obtain the grayscale image of the adhesive surface; in the grayscale image, each pixel point has a corresponding grayscale value, and the edge is the area where the grayscale value changes greatly in the image. By calculating the change of the grayscale value of each pixel point and its neighboring pixel points, the edge intensity of the pixel point is obtained. For example, the Sobel operator calculates the gradients in the horizontal and vertical directions respectively by performing a convolution operation on the image, and then calculates the total edge intensity according to the Pythagorean theorem. For each pixel point in the grayscale image of the adhesive surface, a value representing its edge intensity can be obtained, thus forming an edge intensity distribution.

[0081] Specifically, the preset edge threshold is a user-defined value used to distinguish edge points from non-edge points. The edge intensity of each pixel point is compared with the edge threshold. If the edge intensity is greater than or equal to the edge threshold, the pixel point is considered an edge point; if the edge intensity is less than the edge threshold, the pixel point is determined to be a non-edge point. Thus, all pixel points in the grayscale image of the adhesive surface are divided into two categories: edge points and non-edge points. After identifying the edge points and non-edge points, all the edge points are connected to obtain the global edge of the surface image of the patch adhesive. For example, the edge tracking algorithm starts from an edge point and searches for adjacent edge points according to certain rules, gradually connecting all the edge points into a complete edge contour. The global edge represents the overall boundary information of the patch adhesive in the image.

[0082] Furthermore, although the global edge division can determine the overall boundary of the patch adhesive, it may ignore some subtle features and details. Therefore, it is necessary to perform a global detail scaling operation on the details to improve the image quality.

[0083] In the embodiment of the present invention, the global adhesive surface image refers to the image obtained after processing the surface image of the patch adhesive after global edge division, which more clearly shows the surface features of the patch adhesive, especially the details of the edge part, as well as information such as the abnormal areas highlighted after processing.

[0084] In the embodiment of the present invention, the global detail scaling operation is performed on the surface image of the patch adhesive after global edge division to obtain the global adhesive surface image, including:

[0085] Extract the global edge corresponding to the surface image of the patch adhesive after global edge division;

[0086] Identify the initial edge points of the global edge, and calculate the dynamic edge point coefficient of the global edge according to the initial edge points: Among them, is the dynamic edge point coefficient, is the gray-scale change weight, is the edge change weight, is the initial edge point the number of pixel points in the neighborhood, is the pixel point in the neighborhood gray-scale value, is the average value of the gray-scale of the pixel points in the neighborhood, is the sine function, is the arctangent function, is the target candidate edge point;

[0087] Determine the global edge points of the global edge according to the dynamic edge point coefficient;

[0088] Perform local magnification operations on the global edge points one by one, identify the abnormal images of the patch adhesive surface image according to the magnified global edge points, and perform multi-dimensional enhancement processing on the abnormal images to obtain the edge surface images corresponding to each global edge point;

[0089] Perform shrinkage processing on the edge surface images, identify the edge intensity distribution of the edge images after shrinkage processing, and perform multi-dimensional enhancement processing on the edge images according to the edge intensity distribution to obtain the global adhesive surface image.

[0090] Specifically, the edge contour obtained by the edge detection algorithm is separated from the original image to form an image representation containing only the edge. On the extracted global edge, some starting edge points are determined as the basis for calculation. The initial edge point can be any point on the edge contour, thereby determining the dynamic edge point coefficient. The dynamic edge point coefficient refers to the dynamic interval between two edge points. Let the initial edge point be , and the next candidate edge point be . In the local neighborhood centered on (for example neighborhood), calculate the gray-scale variance . The calculation formula of the gray-scale variance is: , where is the gray-scale value of the point in the neighborhood, is the average value of the gray-scale in the neighborhood, is the number of pixel points in the neighborhood. At the same time, calculate and the direction angle of the line connecting the two points, let the vector , then (the quadrant situation needs to be considered), while and are weight coefficients , the gray variance reflects the degree of gray change in the local area. The greater the degree of change may mean that the edge is more complex and requires a smaller interval to accurately describe; while | reflects the edge orientation. When the edge orientation is significantly different from the horizontal or vertical direction, the interval is adjusted appropriately to adapt to the edge change.

[0091] Specifically, according to the calculated dynamic edge point coefficients of each initial edge point, a certain threshold or rule is set to screen out the points with larger dynamic edge point coefficients or meeting specific conditions as global edge points. The global edge points are considered to be the points with more important features on the edge or better representing the edge morphology; and for each determined global edge point, the local area around it is magnified with this point as the center, that is, realized through an image scaling algorithm, and the image of the local area is magnified to an appropriate ratio to observe details more clearly; in the magnified local image, through image analysis algorithms (such as threshold segmentation, feature matching, etc.), the areas different from the normal situation are identified, that is, abnormal images, such as the detected edge discontinuity, the appearance of additional spots or texture abnormalities, etc.; for the identified abnormal images, enhancement processing is performed from multiple dimensions (such as gray level, contrast, texture, etc.), and methods such as histogram equalization and filtering can be used to improve the visibility of the abnormal area and highlight its features, and finally the edge surface image corresponding to each global edge point is obtained; the enhanced edge surface image is subjected to an appropriate shrinking operation, that is, through downsampling or other image compression algorithms, to reduce the size of the image while retaining the main edge features; calculate the edge intensity distribution of the shrunk image, that is, determine the edge intensity magnitudes at different positions in the image, and then, according to the edge intensity distribution, enhance the edge image from multiple dimensions again to highlight the edge features and obtain the global adhesive surface image.

[0092] Furthermore, although the global adhesive surface image has undergone a series of processes, some local detailed features may be overlooked in the overall observation. Therefore, it is necessary to enhance the image quality inside the global adhesive surface image to more comprehensively understand the condition of the adhesive surface.

[0093] S2. Perform local center division on the global adhesive surface image, perform local detail scaling operation on the globally adhesive surface image after local center division to obtain a local adhesive surface image, and fuse the global adhesive surface image and the local adhesive surface image to obtain a patch adhesive image.

[0094] In the embodiment of the present invention, the local center division refers to dividing the global adhesive surface image into multiple smaller local areas, and the local center division helps to highlight the local features of the image.

[0095] In the embodiments of the present invention, the local center division of the global back glue surface image includes:

[0096] Determine an initial local division window according to the image shape of the global back glue surface image;

[0097] Determine the division area line of the global back glue surface image according to the initial local division window;

[0098] Identify the two-way dynamic window factor of the initial local division window through the division area line;

[0099] Dynamically update the initial local division window using the two-way dynamic window factor, and return to the step of determining the division area line of the global back glue surface image according to the initial local division window until the global back glue surface image is completely divided;

[0100] When the global back glue surface image is completely divided, determine the local area of the global back glue surface image according to each division area line.

[0101] Specifically, the global back glue surface image may have different shapes, such as rectangles, irregular polygons, etc. An initial local division window is determined according to its shape, and the size and shape of the window should be able to reasonably cover a part of the image area. If the image is rectangular, a smaller rectangular window can be selected as the initial division window. After determining the initial local division window, the division area line is determined based on the initial division window. The division area line is the boundary line that divides the image into different local areas, and the division area line can be determined according to the gray scale change or edge information of the image within the window, so that the divided areas can better reflect the local characteristics of the image.

[0102] Specifically, the bidirectional dynamic window factor is used to adjust the parameters of the initial local division window, including the adjustment of the window in the horizontal and vertical directions, such as the change of window size, position movement, etc. If it is found that the image features near the division area line are relatively complex, the window size may need to be increased. Then the bidirectional dynamic window factor will adjust the window size in the horizontal and vertical directions accordingly, and then update the initial local division window according to the calculated bidirectional dynamic window factor to change its size, position or shape. Based on the updated window, the division area line is determined again, and then the bidirectional dynamic window factor is calculated through the new division area line. This process is repeated until the entire global adhesive surface image is divided into multiple local areas, that is, the image is divided completely. When the image is divided, the area enclosed by each division area line is a local area, and the local area is a subdivision of the global adhesive surface image, and each area can be analyzed and processed independently.

[0103] Furthermore, after the local center division, each local area has its corresponding position and range information, and the appropriate zoom ratio is determined according to the position and range information. If you need to observe very subtle details, you can choose a larger zoom ratio, such as 2 times, 5 times or even higher; if you just want to roughly view the overall situation of the local area, a smaller zoom ratio (such as 1.5 times) is used, and then the corresponding local area image is extracted from the global adhesive surface image, and the local area is separated from the original image by image cropping, and the scaled local image is put back to the position of the original image, and then an image fusion operation is performed, and the scaled local image is recombined with other parts of the original image by image stitching. During the scaling process, the local image may exceed the boundary of the original image. At this time, the excess part needs to be filled to avoid blank areas. The filling methods include but are not limited to zero filling (filling the excess part with black) and mirror filling (mirroring the boundary pixels), so as to obtain a local adhesive surface image after the local detail scaling operation, which can more clearly display the detailed features of the local area.

[0104] Furthermore, when analyzing the adhesive on the back of the patch, relying solely on the global image may ignore some important local features, while focusing only on the local image may lose the grasp of the overall situation. By fusing the global and local images, local details can be observed in the overall context, avoiding misjudgment caused by observing the local details in isolation.

[0105] In an embodiment of the present invention, the patch adhesive backing image refers to an image obtained by processing and fusing the global adhesive backing surface image and the local adhesive backing surface image, which includes both the overall macro information of the patch adhesive backing, such as the overall shape, contour, etc., and the local micro detail information.

[0106] In an embodiment of the present invention, the fusion of the global adhesive surface image and the local adhesive surface image to obtain a patch adhesive image includes:

[0107] Identifying local edge features of the local adhesive surface image and global edge features of the global adhesive surface image;

[0108] Comparing the feature positions of the local edge features with those of the global edge features to obtain identical feature positions;

[0109] Performing feature fusion of the local edge features and the global edge features according to the identical feature positions to obtain a patch adhesive image.

[0110] Specifically, edge detection algorithms (such as the Canny algorithm, Sobel operator, etc.) are used to identify their edge information. The edge information reflects features such as the boundaries and shape changes of the adhesive surface within a local area. For the global adhesive surface image, the corresponding edge detection algorithm is used to extract its overall edge information, that is, the global edge features. For example, some tiny crack or texture change edges may be detected in the local adhesive surface image, while the overall contour edge of the adhesive can be identified in the global adhesive surface image.

[0111] Specifically, by comparing the position information of the local edge features and the global edge features, feature points or feature regions with the same or similar positions in the local image and the global image are screened. These positions are called identical feature positions. For example, a feature point at a corner of the adhesive edge in the local image can also be found at the corresponding position in the global image, that is, the identical feature position, so as to ensure the accuracy of the fusion. According to the previously determined identical feature positions, the local edge features and the global edge features are merged or fused. During the fusion process, for edge pixels at the same position, the fused pixel value can be determined according to the fusion rule (such as taking the average value, selecting a clearer edge, etc.), so as to organically combine the local detailed features with the global overall features and obtain an image that combines local and global information, that is, a patch adhesive image.

[0112] Furthermore, since the surface of the patch adhesive may be uneven, and for adhesives with uneven surfaces, undulations, and uneven thicknesses, in order to achieve more accurate cutting positioning, effectively avoid two-dimensional positioning deviations caused by uneven adhesive surfaces, and improve cutting accuracy, it is necessary to identify the three-dimensional information of the patch adhesive surface.

[0113] S3. Using a preset dual-vision positioning algorithm to identify the three-dimensional information of the patch adhesive image, and determining the cutting plane of the patch adhesive image according to the three-dimensional information.

[0114] In the embodiments of the present invention, the three-dimensional information refers to the information describing the position of the target point of the patch back glue in the three-dimensional space. Determining the specific position of the target point in the three-dimensional space can comprehensively and accurately reflect the spatial shape and position characteristics of the patch back glue.

[0115] In the embodiments of the present invention, the recognition of the three-dimensional information of the patch back glue image by using a preset dual-vision positioning algorithm includes:

[0116] Recognizing the shape complexity index and the feature stability index of the patch back glue image;

[0117] Converting the world coordinates of the patch back glue image into image plane coordinates;

[0118] Determining a first weight and a second weight according to the shape complexity index and the feature stability index, where the first weight and the second weight are: Wherein, is the first weight, is the second weight, is the shape complexity index, is the stability index of the machine vision feature, is the stability index of the laser sensor feature;

[0119] Performing three-dimensional fusion on the image plane coordinates and the pre-acquired image measurement height by using the following dual-vision positioning algorithm to obtain the three-dimensional information of the patch back glue image: Wherein, is the abscissa information in the three-dimensional information, is the ordinate information in the three-dimensional information, is the vertical coordinate information in the three-dimensional information, is the first weight, is the second weight, is the plane abscissa of the patch back glue target point in the world coordinate system, is the plane ordinate of the patch back glue target point in the world coordinate system, is the height coordinate of the patch back glue target point, is the projection abscissa of the laser measurement point on the plane, is the projection ordinate of the laser measurement point on the plane, is the target height corresponding to the laser measurement point.

[0120] Specifically, for an image region, the shape complexity is measured by calculating the fractal dimension of its contour, and the fractal dimension can be calculated using the box-counting method. Assume that the contour of the image region is covered with small boxes of side length , and the number of small boxes required is , then the fractal dimension It can be approximately calculated by the following formula: , the higher the shape complexity, the more the fractal dimension deviates from an integer. For example, for a complex and irregular adhesive shape on the patch, its fractal dimension may be between 1.5 and 2, while a regular shape is close to 1 (such as a straight line) or 2 (such as a square). Among them, the machine vision system is responsible for obtaining the planar features and position information of the adhesive on the patch, and the laser displacement sensor accurately measures the three-dimensional information such as the height of the adhesive. For the planar features extracted by the machine vision, the feature stability can be measured by calculating the matching consistency of the feature points. Assume that feature extraction is performed on the same area at different times or from different perspectives, and the feature point sets and are obtained. The number of pairs of matching feature points is , and the total number of feature points is , then the feature matching consistency is . The higher the consistency, the more stable the feature. For the height data measured by the laser displacement sensor, the stability can be measured by calculating the standard deviation of multiple measurements. Assume that height measurements are made at a certain point, and the measured values are . Then calculate the standard deviation corresponding to the measured values. The smaller the standard deviation, the more stable the height measurement. Then, based on the shape complexity and the feature stability (machine vision feature), (laser displacement sensor feature) to adjust the weight coefficient. When the shape complexity is relatively low (the shape is relatively regular) and the stability of the machine vision feature is relatively high, is relatively large, that is, increase the weight of the machine vision measurement data. When the shape complexity is relatively high and the stability of the laser displacement sensor feature is relatively high, is relatively large, that is, increase the weight of the laser displacement sensor measurement data.

[0121] Specifically, in the machine vision system, it is usually necessary to convert the three-dimensional coordinates of an object in the world coordinate system into two-dimensional coordinates on the image plane. Then, based on the imaging model of the camera, such as the pinhole camera model, it is realized through the internal parameters of the camera (focal length , image center coordinates ) and external parameters (rotation matrix and translation vector ). The rotation matrix of the camera is a 3×3 matrix that describes the rotation relationship of the camera coordinate system relative to the world coordinate system. The translation vector of the camera is a 3×1 vector, representing the position of the origin of the camera coordinate system in the world coordinate system. It is the coordinate estimated from the projection of the laser measurement points on the plane and participates in the fusion as auxiliary information. It is the height estimated from the machine vision image features and is also used as auxiliary information for fusion.

[0122] Furthermore, the cutting plane of the patch adhesive image is determined according to the three-dimensional information. That is, the known three-dimensional information consists of the abscissa , the ordinate and the vertical coordinate which together determine the positions of each point on the patch adhesive in three-dimensional space, thus being able to comprehensively reflect the spatial shape of the patch adhesive. Then, the plane determined by each vertical coordinate in the three-dimensional coordinates is used as a cutting plane. On this plane, the abscissa and the ordinate can take various values, while the vertical coordinate always remains , thereby determining the shape and characteristics of the patch adhesive at different heights.

[0123] Even further, the cutting plane is a plane fixed by the vertical coordinate. On this plane, the abscissa and the ordinate are two independent and orthogonal dimensions that can fully describe the position of any point on the plane. Therefore, cutting along these two directions can flexibly cut objects on the plane into various shapes and sizes to meet different cutting requirements.

[0124] S4. Identify the two-way cutting direction corresponding to the cutting plane, and generate a two-way cutting path for the patch adhesive image according to the two-way cutting direction and the cutting plane.

[0125] In the embodiments of the present invention, the two-way cutting direction refers to cutting along the two mutually perpendicular directions of the abscissa and the ordinate within the cutting plane. Among them, since on the cutting plane (the plane with the vertical coordinate fixed as ) the abscissa can take various values, cutting can be performed along the abscissa direction. The direction from left to right or from right to left is a cutting direction, and on this direction, the starting point, ending point, and path of cutting can be determined as needed to achieve cutting of the patch adhesive in this direction. For example, the patch adhesive can be cut into parts with different widths in this direction; similarly, on this cutting plane, the ordinate can also take different values, so cutting can also be performed along the ordinate direction, which can be the direction from top to bottom or from bottom to top. Through cutting in this direction, the patch adhesive can be segmented or trimmed in this dimension. For example, the patch adhesive can be cut along the ordinate direction according to specific length requirements.

[0126] Furthermore, the two-way cutting directions are perpendicular to each other, which can provide more accurate positioning and control. In order to reduce the idle stroke and repeated cutting during the cutting process and improve the cutting efficiency, it is necessary to reasonably plan the two-way cutting path.

[0127] In the embodiment of the present invention, the two-way cutting path refers to a path set formed by combining cutting paths generated along two mutually perpendicular cutting directions within the cutting plane, which clarifies the specific routes and sequences of cutting the patch backing glue in two directions, enabling the cutting operation to comprehensively and accurately cover the area to be cut, thereby achieving effective cutting of the patch backing glue.

[0128] In the embodiment of the present invention, generating the two-way cutting path of the patch backing glue image according to the two-way cutting direction and the cutting plane includes:

[0129] Arbitrarily select the cutting plane as the initial cutting plane, and gradually move from any endpoint in the initial cutting plane according to the two-way cutting direction to obtain an initial cutting path;

[0130] Extend the initial cutting path according to the first cutting direction in the two-way cutting direction and the cutting sequence of the cutting plane to obtain a first cutting path;

[0131] Extend the initial cutting path according to the second cutting direction in the two-way cutting direction and the cutting sequence to obtain a second cutting path;

[0132] Generate the two-way cutting path of the patch backing glue image according to the first cutting path and the second cutting path.

[0133] Specifically, arbitrarily select one of all the cutting planes determined by the vertical coordinate as the starting cutting plane. Based on the two-way cutting direction (the horizontal coordinate direction and the vertical coordinate direction), start from any endpoint (such as the upper left corner, the lower right corner, etc.) of this initial cutting plane, and gradually move according to the two-way cutting direction, moving a small unit distance each time (for example, the distance of one pixel point), and connect the points passed by the movement in sequence to form an initial cutting path. Then the initial cutting path is the starting point for extending and improving the cutting path.

[0134] Specifically, the first cutting direction is the abscissa direction. According to the cutting sequence of the cutting plane (which can be preset, such as from left to right, from top to bottom, etc.), along the first cutting direction, on the basis of the initial cutting path, the path is continuously extended. The extension method is also to move step by step according to certain rules, moving a suitable distance each time, and connecting the newly passed points to the initial cutting path, thus obtaining a longer path, that is, the first cutting path; when the first cutting direction is the ordinate direction, according to the cutting sequence of the cutting plane, the initial cutting path is extended along the second cutting direction. In the same way of step-by-step movement and connecting points, the second cutting path is obtained. The second cutting path cuts the patch back glue in the second cutting direction, cooperates with the first cutting path, forms a cutting coverage of the patch back glue in two directions, and then combines the first cutting path and the second cutting path to obtain a two-way cutting path of the patch back glue image. The two-way cutting path contains cutting information in two mutually perpendicular directions, and can guide the cutting device to perform a complete two-way cutting operation on the patch back glue within the cutting plane to achieve the expected cutting effect.

[0135] Furthermore, the two-way cutting path describes the specific routes for cutting the patch back glue in two mutually perpendicular directions. In order to clarify information such as the starting point, ending point of different cutting paths and their intersection positions within the cutting plane, it is necessary to display the cutting path in a graphical way.

[0136] S5. Generate a cutting connection diagram of the patch back glue image according to the two-way cutting path, and locate the cutting position of the patch back glue image through the cutting connection diagram.

[0137] In the embodiment of the present invention, the cutting connection diagram refers to a graph formed by discretizing the two-way cutting path of the patch back glue image, identifying the intersection points in the path discrete points and determining the connection relationship, and then connecting the path discrete points according to the connection relationship.

[0138] In the embodiment of the present invention, the generating the cutting connection diagram of the patch back glue image according to the two-way cutting path includes:

[0139] Discretize the two-way cutting path to obtain path discrete points;

[0140] Identify the intersection points in the path discrete points, and determine the connection relationship between the path discrete points according to the positions corresponding to the intersection points;

[0141] Connect the path discrete points according to the connection relationship to obtain a cutting connection diagram.

[0142] Specifically, discretization is to divide a continuous path into a series of discrete points. The points can be selected along the path at a certain interval (such as a fixed distance or the number of pixels). For example, in a graphical cutting path, a point is marked every certain length, then the original continuous cutting path is represented as a set of discrete points, that is, path discrete points. After obtaining the path discrete points, it is necessary to find the intersection points among them, that is, the points where two or more cutting paths intersect. Once the intersection points are identified, the connection method can be determined according to the positions of the intersection points and the relative position relationships between each path discrete point and the intersection points. For example, if two path discrete points are adjacent to the same intersection point and their orders on their respective paths are continuous, then there is a connection relationship between these two points, so as to determine the mutual connection situation among all the discrete points in the entire cutting path. Furthermore, after connecting all the points that meet the connection relationship in sequence, a cutting connection diagram is obtained. The cutting connection diagram visually shows the structure of the bidirectional cutting path and the connection situation between each part. For example, in a simple cutting path, there are multiple discrete points and several intersection points. After connecting these points according to the connection relationship, the trend of the cutting path and the connection order between each path segment can be clearly seen.

[0143] Furthermore, the cutting connection diagram clearly shows the cutting path of the entire patch adhesive image in a graphical manner, converting the complex cutting path into intuitive lines and nodes. Then, through the cutting connection diagram, the starting position, ending position, and connection points of each path segment can be clearly distinguished, so as to accurately determine the cutting position.

[0144] In the embodiment of the present invention, the cutting position refers to the specific position point or position area on the surface of the patch adhesive that needs to be cut according to the cutting requirements and the cutting connection diagram.

[0145] In the embodiment of the present invention, positioning the cutting position of the patch adhesive image through the cutting connection diagram includes:

[0146] Identifying the cutting density of the patch adhesive image according to the cutting connection diagram;

[0147] Determining the cutting area of the patch adhesive image according to the cutting density;

[0148] Determining the cutting position of the patch adhesive image through the cutting area.

[0149] Specifically, the cutting density refers to the density of cutting paths in a certain area of the patch adhesive image. For example, in the cutting connection diagram, if there are more cutting path segments and denser distribution of path discrete points in a certain area, it indicates a higher cutting density in that area; conversely, if there are fewer cutting path segments and sparser path discrete points in a certain area, the cutting density is lower. By analyzing the distribution of paths in the cutting connection diagram, the cutting density of different areas can be calculated or visually judged. For example, the number of path segments or the number of path discrete points per unit area can be counted to quantify the cutting density. After determining the cutting density, the cutting area can be divided according to certain criteria or requirements. That is, the area with a higher cutting density may be the part that requires more precise or more cutting operations, while the area with a lower cutting density requires relatively fewer cutting operations. For example, a cutting density threshold is set, and the area with a cutting density higher than this threshold is divided into one cutting area, and the area lower than this threshold is divided into another or multiple cutting areas. In a cutting area, the starting point, ending point, and key turning points on the path are all important cutting positions. By analyzing the coordinate information of these positions in the cutting connection diagram, the cutting positions on the patch adhesive image can be accurately located.

[0150] Exemplarily, in the cutting area, the starting point coordinates are (10, 10), indicating that in the plane coordinate system of the patch adhesive image, this cutting path starts from the position with an abscissa of 10 mm and an ordinate of 10 mm. There are two key turning points, with coordinates (30, 20) and (50, 15) respectively. The first turning point indicates that the cutting path changes direction when it reaches the position with an abscissa of 30 mm and an ordinate of 20 mm; the second turning point is where the cutting path changes direction again at the position with an abscissa of 50 mm and an ordinate of 15 mm. The ending point coordinates are (70, 30), that is, when the cutting path reaches the position with an abscissa of 70 mm and an ordinate of 30 mm, the cutting in this cutting area ends. Then, the cutting head is first moved to the starting point (10, 10) to start cutting. When cutting to the key turning point (30, 20), the cutting direction is changed, and cutting continues to (50, 15) where the direction is changed again, and finally cutting reaches the ending point (70, 30) to complete this section of the cutting operation in this cutting area.

[0151] As Figure 2 shown, it is the functional module diagram of the high-precision cutting positioning system for patch adhesive based on machine vision provided by an embodiment of the present invention.

[0152] The high-precision cutting and positioning system 100 for patch back glue based on machine vision according to the present invention can be installed in an electronic device. According to the functions achieved, the high-precision cutting and positioning system 100 for patch back glue based on machine vision can include an image division module 101, an image fusion module 102, a cutting plane determination module 103, a bidirectional cutting path generation module 104, and a cutting position positioning module 105. The modules in the present invention can also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0153] In this embodiment, the functions of each module / unit are as follows:

[0154] The image division module 101 is configured to collect the surface image of the patch back glue based on a preset all-round angle of image acquisition, perform global edge division on the surface image of the patch back glue, and perform a global detail scaling operation on the surface image of the patch back glue after global edge division to obtain a global back glue surface image;

[0155] The image fusion module 102 is configured to perform local center division on the global back glue surface image, perform a local detail scaling operation on the global back glue surface image after local center division to obtain a local back glue surface image, and fuse the global back glue surface image and the local back glue surface image to obtain a patch back glue image;

[0156] The cutting plane determination module 103 is configured to identify the three-dimensional information of the patch back glue image by using a preset dual-vision positioning algorithm, and determine the cutting plane of the patch back glue image according to the three-dimensional information;

[0157] The bidirectional cutting path generation module 104 is configured to identify the bidirectional cutting direction corresponding to the cutting plane, and generate a bidirectional cutting path of the patch back glue image according to the bidirectional cutting direction and the cutting plane;

[0158] The cutting position positioning module 105 is configured to generate a cutting connection diagram of the patch back glue image according to the bidirectional cutting path, and locate the cutting position of the patch back glue image through the cutting connection diagram.

[0159] Specifically, each module in the high-precision cutting and positioning system 100 for patch back glue based on machine vision in the embodiment of the present invention uses the same technical means as those Figure 1 described in the high-precision cutting and positioning method for patch back glue based on machine vision described above, and can produce the same technical effects, which will not be elaborated here.

[0160] In several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation.

[0161] The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0162] In addition, in each embodiment of the present invention, the functional modules can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a combination of hardware and software functional modules.

[0163] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0164] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is not limited only by the above description. Therefore, it is intended to include all changes within the meaning and scope of equivalent elements falling within the protection scope in the present invention.

[0165] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is a theory, method, technology and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0166] In addition, obviously, the word "including" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or systems stated in the system claims can also be implemented by one unit or system through software or hardware. The words such as first and second are used to represent names and do not represent any specific order.

[0167] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A high-precision cutting and positioning method for patch back glue based on machine vision, characterized in that, The method includes: Collecting the surface image of the patch adhesive based on a preset all-round image acquisition angle, performing global edge division on the surface image of the patch adhesive, and performing a global detail scaling operation on the surface image of the patch adhesive after global edge division to obtain a global adhesive surface image; Performing local center division on the global adhesive surface image, performing a local detail scaling operation on the global adhesive surface image after local center division to obtain a local adhesive surface image, and fusing the global adhesive surface image and the local adhesive surface image to obtain a patch adhesive image; Identifying the three-dimensional information of the patch adhesive image by using a preset dual-vision positioning algorithm, and determining the cutting plane of the patch adhesive image according to the three-dimensional information; Identifying the two-way cutting direction corresponding to the cutting plane, and generating a two-way cutting path of the patch adhesive image according to the two-way cutting direction and the cutting plane; Generating a cutting connection diagram of the patch adhesive image according to the two-way cutting path, and positioning the cutting position of the patch adhesive image through the cutting connection diagram.

2. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, wherein The collecting the surface image of the patch adhesive based on a preset all-round image acquisition angle includes: Extracting the horizontal direction and the vertical direction corresponding to the all-round image acquisition angle; Dividing the all-round image acquisition angle into horizontal all-round angles according to a preset horizontal angle interval in the horizontal direction; Dividing the all-round image acquisition angle into vertical all-round angles according to a preset vertical angle interval in the vertical direction; Configuring all-round image acquisition points according to the horizontal all-round angles and the vertical all-round angles, and collecting the surface image of the patch adhesive by using the all-round image acquisition points.

3. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, characterized in that, The performing global edge division on the surface image of the patch adhesive includes: Converting the surface image of the patch adhesive into an adhesive surface grayscale image; Identifying the pixel points of the adhesive surface grayscale image, and generating an edge intensity distribution of the adhesive surface grayscale image according to the pixel points; Identifying edge points and non-edge points by using the edge intensity distribution and a preset edge threshold; Determining the global edge of the surface image of the patch adhesive according to the edge points and the non-edge points.

4. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, characterized in that, The performing a global detail scaling operation on the surface image of the patch adhesive after global edge division to obtain a global adhesive surface image includes: Extracting the global edge corresponding to the surface image of the patch adhesive after global edge division; Identifying the initial edge points of the global edge, and calculating the dynamic edge point coefficient of the global edge according to the initial edge points; Among them, C is the dynamic edge point coefficient, α is the gray-scale change weight, β is the edge change weight, |N(x0, y0)| is the number of pixel points in the neighborhood of the initial edge point (x0, y0), I(i, j) is the gray-scale value of the pixel point (i, j) in the neighborhood, is the average value of the gray-scale of the pixel points in the neighborhood, sin is the sine function, arctan is the arctangent function, and (x1, y1) is the target candidate edge point; Determining the global edge points of the global edge according to the dynamic edge point coefficient; Performing a local magnification operation on each of the global edge points one by one, identifying abnormal images of the surface image of the patch adhesive according to the magnified global edge points, and performing multi-dimensional enhancement processing on the abnormal images to obtain an edge surface image corresponding to each global edge point; Performing a shrinking process on the edge surface image, identifying the edge intensity distribution of the edge image after the shrinking process, and performing multi-dimensional enhancement processing on the edge image according to the edge intensity distribution to obtain a global adhesive surface image.

5. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, characterized in that The local center division of the global adhesive surface image includes: Determining an initial local division window according to the image shape of the global adhesive surface image; Determining the division area line of the global adhesive surface image according to the initial local division window; Identifying the two-way dynamic window factor of the initial local division window through the division area line; Dynamically updating the initial local division window by using the two-way dynamic window factor, and returning to the step of determining the division area line of the global adhesive surface image according to the initial local division window until the global adhesive surface image is completely divided; When the global adhesive surface image is completely divided, determining the local area of the global adhesive surface image according to each division area line.

6. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, characterized in that The fusion of the global adhesive surface image and the local adhesive surface image to obtain a patch adhesive image includes: Identifying the local edge features of the local adhesive surface image and the global edge features of the global adhesive surface image; Comparing the feature positions of the local edge features with the feature positions of the global edge features to obtain the same feature positions; Performing feature fusion on the local edge features and the global edge features according to the same feature positions to obtain a patch adhesive image.

7. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, characterized in that The use of a preset dual vision positioning algorithm to identify the three-dimensional information of the patch adhesive image includes: Identifying the shape complexity index and the feature stability index of the patch adhesive image; Converting the world coordinates of the patch adhesive image into image plane coordinates; Determining a first weight and a second weight according to the shape complexity index and the feature stability index, where the first weight and the second weight are: Where ω1 is the first weight, ω2 is the second weight, S is the shape complexity index, B1 is the stability index of the machine vision feature, and B2 is the stability index of the laser sensor feature; Using the following dual vision positioning algorithm to perform three-dimensional fusion on the image plane coordinates and the pre-acquired image measurement height to obtain the three-dimensional information of the patch adhesive image: Where X is the abscissa information in the three-dimensional information, Y is the ordinate information in the three-dimensional information, Z is the vertical coordinate information in the three-dimensional information, ω1 is the first weight, ω2 is the second weight, X1 is the plane abscissa of the patch adhesive target point in the world coordinate system, Y1 is the plane ordinate of the patch adhesive target point in the world coordinate system, Z1 is the height coordinate of the patch adhesive target point, X2 is the projection abscissa of the laser measurement point on the plane, Y2 is the projection ordinate of the laser measurement point on the plane, and Z2 is the target height corresponding to the laser measurement point.

8. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, wherein The generation of the two-way cutting path of the patch adhesive image according to the two-way cutting direction and the cutting plane includes: Randomly selecting the cutting plane as the initial cutting plane, and gradually moving from any endpoint in the initial cutting plane according to the two-way cutting direction to obtain an initial cutting path; Extending the initial cutting path according to the first cutting direction in the two-way cutting direction and the cutting sequence of the cutting plane to obtain a first cutting path; Extend the initial cutting path according to the second cutting direction in the described two-way cutting direction and the cutting sequence to obtain a second cutting path; Generate a two-way cutting path of the patch back glue image according to the first cutting path and the second cutting path.

9. The high-precision cutting and positioning method for patch back glue based on machine vision according to claim 1, wherein The generation of the cutting connection diagram of the patch back glue image according to the two-way cutting path includes: Discretize the two-way cutting path to obtain path discrete points; Identify the intersection points among the path discrete points, and determine the connection relationship between the path discrete points according to the positions corresponding to the intersection points; Connect the path discrete points according to the connection relationship to obtain a cutting connection diagram.

10. A high-precision cutting and positioning system for patch back glue based on machine vision, characterized in that, For implementing the high-precision cutting and positioning method of patch back glue based on machine vision according to any one of claims 1-9, the system includes: An image division module, configured to collect a surface image of the patch back glue based on a preset all-round angle of image acquisition, perform global edge division on the surface image of the patch back glue, and perform a global detail scaling operation on the surface image of the patch back glue after global edge division to obtain a global back glue surface image; An image fusion module, configured to perform local center division on the global back glue surface image, perform a local detail scaling operation on the global back glue surface image after local center division to obtain a local back glue surface image, and fuse the global back glue surface image and the local back glue surface image to obtain a patch back glue image; A cutting plane determination module, configured to identify three-dimensional information of the patch back glue image by using a preset dual vision positioning algorithm, and determine a cutting plane of the patch back glue image according to the three-dimensional information; A two-way cutting path generation module, configured to identify a two-way cutting direction corresponding to the cutting plane, and generate a two-way cutting path of the patch back glue image according to the two-way cutting direction and the cutting plane; A cutting position positioning module, configured to generate a cutting connection diagram of the patch back glue image according to the two-way cutting path, and position the cutting position of the patch back glue image through the cutting connection diagram.

Images