A glue coating method suitable for large-size glass
By analyzing point cloud data and optimizing edge feature factors of large-size glass, the problem of adhesive coating path planning was solved, achieving uniformity and continuity in the adhesive coating process, and improving adhesive coating quality and production efficiency.
Patent Information
- Application Number
- CN202410939770.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-07-15
AI Technical Summary
In the process of applying adhesive to large-size glass, the path planning of the adhesive application robot is difficult, the unevenness of the glass surface affects the uniformity of adhesive application, and the existing technology is not precise enough, resulting in adhesive application quality problems.
By acquiring point cloud data of the glass, marking the contour point cloud data and mapping it to a two-dimensional plane, analyzing the local features and unevenness influencing factors of pixels, extracting edge feature factors, and optimizing the adhesive application path.
It enables precise quantification and evaluation of glass surface flatness, ensuring the accuracy and efficiency of the adhesive application path, guaranteeing adhesive quality and continuity, reducing adhesive waste, and improving production efficiency.
Smart Images

Figure CN118469118B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image processing technology, and in particular to a gluing method suitable for large-size glass. Background Art
[0002] The glue coating process of large-size glass plays a vital role in semiconductor chip manufacturing and liquid crystal display production. The development of 3D real-time glue coating detection technology provides a more accurate and efficient solution for glue coating quality monitoring, which helps to promote technological progress in the entire industry.
[0003] In the automated gluing production line, the gluing robot needs to apply glue precisely on the glass surface to ensure the sealing and structural strength of the product. However, due to the increase in glass size and the complexity of its shape, the path planning of the gluing robot becomes very difficult. In addition, the slight unevenness of the glass surface will also affect the uniformity of the gluing. If the gluing path planning is unreasonable, it will not only lead to waste of glue, but also may cause uneven gluing, affecting the final quality of the product. The existing gluing robots are not accurate enough in analyzing the geometric shape of the glass edge and evaluating the surface flatness. The path point selection during gluing is greatly affected by the surface flatness, resulting in problems with the quality of the glass gluing effect. Summary of the invention
[0004] In view of the above, it is necessary to provide a gluing method suitable for large-size glass, plan the gluing path according to the flatness of the glass surface, and ensure the uniformity and continuity of the gluing process.
[0005] An embodiment of the present application provides a method for applying glue to large-sized glass, the method comprising:
[0006] Obtaining point cloud data of the glass to be coated with glue, marking the contour point cloud data; mapping the point cloud data to be coated with glue and the contour point cloud data on a two-dimensional plane as a target image and preliminary pixel points of the target image respectively;
[0007] According to the local pixel value distribution characteristics of all pixels in the target image, the unstable area is obtained; according to the distance characteristics and shape characteristics of each prepared pixel and the corresponding unstable area, the non-flatness influence factor of each prepared pixel is obtained;
[0008] According to the distribution of the prepared pixel points, the edge features of the coated glass are extracted, and the edge influence feature factor of each prepared pixel point is obtained in combination with the non-flatness influence factor;
[0009] All the prepared pixel points are screened according to the edge influence characteristic factors, and the final gluing path is obtained by combining the optimization algorithm.
[0010] In one embodiment, the step of obtaining the unstable area according to the local pixel value distribution characteristics of all pixels in the target image comprises:
[0011] Divide the pixel values of all pixels in the target image into a preset number of levels;
[0012] Taking each pixel point as the center, a preset window is obtained, and the standard deviation of the corresponding levels of all pixel points in the preset window is calculated, which is recorded as the smoothness factor of each pixel point; all pixel points in the target image whose smoothness factor is not zero are recorded as undulating pixels, and a connection graph is obtained as nodes using a priority search algorithm; each area in the connection graph is recorded as an unstable area.
[0013] In one embodiment, the non-flatness influencing factor of each prepared pixel point is obtained according to the distance characteristics and shape characteristics of each prepared pixel point and the unstable area, specifically:
[0014] Obtain the shortest Euclidean distance and the longest Euclidean distance between each prepared pixel and the center point of all unstable areas; and use the unstable area corresponding to the shortest Euclidean distance of each prepared pixel as the unstable area corresponding to each prepared pixel;
[0015] According to the distance distribution and area distribution of each pixel and its corresponding unstable area, the fluctuation importance weight of each prepared pixel is obtained;
[0016] According to the distribution of the undulating pixels in the unstable area corresponding to each prepared pixel point, the regional non-smoothness feature of each prepared pixel point is obtained;
[0017] The product of the regional unevenness feature of each prepared pixel point and the undulation importance weight is used as the unevenness influencing factor of each prepared pixel point.
[0018] In one embodiment, the obtaining of the fluctuation importance weight of each prepared pixel point is specifically as follows:
[0019] The ratio of the area of the unstable region corresponding to each prepared pixel point to the maximum area of all unstable regions is used as the fluctuation regional factor of each prepared pixel point;
[0020] The ratio of the shortest Euclidean distance to the longest Euclidean distance of each prepared pixel point is used as the fluctuation distance factor of each prepared pixel point;
[0021] Obtaining the ups and downs importance weight of each prepared pixel point according to the ups and downs regional factor and the ups and downs distance factor of each prepared pixel point;
[0022] The fluctuation importance weight of each prepared pixel point is in increasing relationship with the fluctuation regional factor and the fluctuation distance factor.
[0023] In one embodiment, the obtaining of the regional non-flat features of each preliminary pixel point is specifically as follows:
[0024] For the non-stationary area corresponding to each prepared pixel point, the average value and variance of the smoothness factors of all the fluctuating pixels are obtained; the square value of the difference between the smoothness factor of each fluctuating pixel point and the average value is calculated, the inverse of the ratio of the square value of the difference to the variance is calculated, and the mapping of the inverse under the exponential function is recorded as the non-smooth characteristic factor of each fluctuating pixel point;
[0025] The cumulative sum of all the non-smooth feature factors in the non-smooth region corresponding to each prepared pixel point is obtained, and recorded as the regional non-smooth feature of each prepared pixel point.
[0026] In one embodiment, the step of obtaining the edge influence characteristic factor of each preliminary pixel point comprises the following steps:
[0027] Corner point detection is performed on the target image, and the characteristic distance of each prepared pixel point is obtained according to the distance distribution between each prepared pixel point and the corner point;
[0028] A local window is preset with each prepared pixel point as the center; wherein the local window is increased by a preset step size and a preset number of increases each time; according to the change of the local window of each prepared pixel point and the change of the shape feature of the edge, the skeleton complex feature factor of each prepared pixel point is obtained;
[0029] Obtaining an edge influence characteristic factor of each prepared pixel point according to the characteristic distance of each prepared pixel point, the skeleton complexity characteristic factor, and the non-flatness influence factor;
[0030] The edge influence characteristic factor of each prepared pixel point is in a consistent trend relationship with the skeleton complexity characteristic factor, the characteristic distance and the non-flatness influence factor.
[0031] In one embodiment, the step of obtaining the characteristic distance of each preliminary pixel point is as follows:
[0032] Corner point detection is performed on the target image to obtain all corner points in the target image, and the shortest Euclidean distance between each prepared pixel point and all corner points is used as the feature distance of each prepared pixel point.
[0033] In one embodiment, the step of obtaining the skeleton complex feature factor of each preliminary pixel point comprises:
[0034] For the local window of each prepared pixel point, the edge skeleton curvature after each growth of the local window is obtained, and the square value of the difference between the edge skeleton curvature of each growth of the local window and the last growth is recorded as the edge skeleton curvature change factor after each growth;
[0035] The density weight after each growth is obtained according to the edge skeleton curvature change factor after each growth and the number of prepared pixels in the corresponding local window;
[0036] The cumulative sum of the product of the density weight and the edge skeleton curvature change factor under all growth times of the local window is calculated as the skeleton complex feature factor of each prepared pixel point.
[0037] In one embodiment, the density weight after each increase is obtained as follows:
[0038] The ratio of the number of prepared pixels after each growth of the local window to the total number of pixels in the local window is used as the edge skeleton density after each growth;
[0039] The product of the edge skeleton density after each growth and the corresponding growth times is calculated, and the mapping of the inverse of the product under the exponential function is used as the density weight.
[0040] In one embodiment, the final gluing path is obtained as follows:
[0041] The prepared pixel points whose edge influence characteristic factors are greater than the preset influence threshold are used as path points during gluing, and the gluing path is obtained by using a random search algorithm based on all the path points.
[0042] This application has at least the following beneficial effects:
[0043] In the above scheme, the point cloud data of the glass to be coated with glue is obtained, and the contour point cloud data is marked to determine the contour of the glue coating, and subsequent analysis is carried out based on the contour features; the point cloud data is mapped to a two-dimensional plane, and the two-dimensional nature of the image is used for visual analysis. In addition, the features and information of the prepared pixel points can be more easily extracted; the unstable area on the glass during the coating process is obtained, and the non-flatness influencing factor is constructed to reflect the microscopic unevenness characteristics of the glass surface, thereby realizing the accurate quantification and evaluation of the flatness of the glass surface; in view of the geometric complexity of the edge of the glass to be coated with glue, the edge influencing characteristic factor is constructed. It reflects the curvature change and edge density characteristics of the edge skeleton, so as to achieve accurate evaluation and optimization of the geometric complexity of the gluing path; by calculating the edge influence characteristic factor of each prepared pixel point and setting a threshold for screening, it can improve the accuracy and efficiency of gluing path selection and ensure that only path points that have a significant impact on the gluing quality are selected; through the selection of path points based on edge influence characteristic factors, the optimal planning of the gluing path is achieved, ensuring the uniformity and continuity of the gluing process, and can complete the gluing of glass of different shapes and sizes and ensure the gluing quality, while reducing the waste of glue materials and improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 A flow chart of a method for gluing large-sized glass provided in this application;
[0045] Figure 2 Schematic diagram of point cloud data collection provided for this application;
[0046] Figure 3 A schematic diagram of the distribution of the prepared pixels provided for this application;
[0047] Figure 4 A flowchart for obtaining the non-flatness influencing factor provided for this application;
[0048] Figure 5 A schematic diagram of corner detection provided for this application;
[0049] Figure 6 A schematic diagram of the process of obtaining the edge influence characteristic factor provided in this application. DETAILED DESCRIPTION
[0050] In the description of the embodiments of the present application, words such as "exemplary", "or", "for example" and the like are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "or", "for example" and the like is intended to present related concepts in a concrete manner.
[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the present application. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application.
[0052] It should also be noted that the terms "first" and "second" in the specification and drawings of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. The method disclosed in the embodiments of the present application or the method shown in the flowchart includes one or more steps for implementing the method. Without departing from the original purpose, the execution order of multiple steps can be interchanged with each other, and some steps can also be deleted.
[0053] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
[0054] The present application embodiment first proposes a method for coating large-size glass, which is applied in the field of image processing technology. Figure 1 , the method comprises the following steps:
[0055] S1: acquiring point cloud data of the glass to be coated with glue, marking the contour point cloud data; mapping the point cloud data to be coated with glue and the contour point cloud data on a two-dimensional plane as a target image and preliminary pixel points of the target image, respectively.
[0056] This application mainly scans the glass to be coated with glue to determine the contour of the glue coating, thereby ensuring the precision and accuracy of the glue coating.
[0057] Based on this, a structured light 3D system is used to obtain the point cloud data of the glass to be coated with glue and mark the contour point cloud data, and the point cloud data of the glass to be coated with glue is mapped into a two-dimensional image to obtain a target image; the points where the contour point cloud data is mapped to the target image are used as preliminary pixel points of the target image.
[0058] It should be noted that using a structured light 3D system to obtain point cloud data of an object and mapping the point cloud data to a two-dimensional plane is a well-known technology and will not be described in detail in this application. Figure 2 As shown; the schematic diagram of the distribution of the prepared pixels is as follows: Figure 3 shown.
[0059] S2: Obtain the unstable area according to the local pixel value distribution characteristics of all pixels in the target image; obtain the non-flatness influence factor of each prepared pixel according to the distance characteristics and shape characteristics between each prepared pixel and the corresponding unstable area.
[0060] For curved glass or irregularly shaped glass, there may be areas that are difficult to cover or uneven glue thickness when applying glue. Therefore, appropriate glue application technology and path planning are needed to solve these problems and ensure the quality of glue application.
[0061] In order to meet the gluing requirements of glass of different specifications and shapes, the pixel values of all pixels in the target image are divided into a preset number of levels. In some embodiments of the present application, the preset number is 64, and the level division adopts the average division method.
[0062] Since the analysis method for each pixel is the same, in some embodiments of the present application, the i-th pixel in the target image is taken as an example for analysis, the i-th pixel is taken as the center, a preset window is obtained, and the standard deviation of the corresponding levels of all pixels in the preset window is calculated, which is recorded as the smoothness factor of the i-th pixel. In some embodiments of the present application, the preset window size corresponding to each pixel is Through the flatness factor, the glue-applying robot can identify tiny uneven areas on the glass surface, facilitating subsequent analysis of the uneven areas.
[0063] All pixels in the target image whose flatness factor is not zero are recorded as undulating pixels, and all undulating pixels are taken as a node, and the connection graph of all nodes is obtained using the breadth-first search algorithm. The area formed by connecting the nodes in the connection graph is recorded as an unstable area. The connection graph can help the glue-coating robot identify the undulating points in the target image, which is helpful for identifying the global features of the glass surface, and the unstable area facilitates the subsequent analysis of local features.
[0064] It should be noted that the breadth-first search algorithm is a well-known technology and will not be described in detail in this application.
[0065] In the embodiment of the present application, the Zhang-Suen edge thinning algorithm is adopted. This algorithm is a well-known technology and will not be described in detail in the present application.
[0066] Since the processing method for each prepared pixel is consistent, the j-th prepared pixel is taken as an example for analysis:
[0067] The shortest Euclidean distance and the longest Euclidean distance between the j-th preparation pixel point and the center point of all the unstable areas are obtained; the unstable area corresponding to the shortest Euclidean distance of each preparation pixel point is used as the unstable area corresponding to each preparation pixel point; the ratio of the area of the unstable area corresponding to each preparation pixel point to the maximum area of all the unstable areas is used as the fluctuation regional factor of the j-th preparation pixel point; the ratio of the shortest Euclidean distance to the longest Euclidean distance is used as the fluctuation distance factor of the j-th preparation pixel point; the fluctuation importance weight of the j-th preparation pixel point is in increasing relationship with the fluctuation regional factor and the fluctuation distance factor.
[0068] It should be noted that an increasing relationship indicates a trend in which as one variable increases, another variable also increases. Common increasing relationships include: linear increasing relationship, exponential increasing relationship, logarithmic increasing relationship, power function increasing relationship, periodic increasing relationship, gradual increasing relationship, etc. This application does not impose any restrictions on this.
[0069] In some embodiments of the present application, the product of the fluctuation regional factor and the fluctuation distance factor of the j-th preliminary pixel point is used as the fluctuation importance weight of the j-th preliminary pixel point. The fluctuation importance weight quantifies the degree of high fluctuation importance of the preliminary pixel point in the nearest unstable region through the relative size and distance of the unstable region closest to the j-th preliminary pixel point.
[0070] For the unsteady region corresponding to the shortest Euclidean distance of the j-th preparation pixel point, the average value and variance of the smoothness factors of all the undulating pixels are obtained; the square value of the difference between the smoothness factor of each undulating pixel point and the average value is calculated, and the inverse of the ratio of the square value of the difference to the variance is calculated, and the mapping of the inverse under the exponential function with the natural constant as the base is recorded as the unsteady characteristic factor of each undulating pixel point; the cumulative sum of all the unsteady characteristic factors in the unsteady region where the j-th preparation pixel point is located is obtained, and recorded as the regional unsteady feature of the j-th preparation pixel point. The regional unsteady feature provides a weighting coefficient relative to the overall average value for the smoothness factor of each undulating pixel point to quantify the flatness feature of the unsteady region closest to the j-th preparation pixel point.
[0071] According to the regional unevenness feature of the jth preliminary pixel and the importance weight of the undulation, the unevenness influence factor of the jth preliminary pixel is obtained; in some embodiments of the present application, the product of the regional unevenness feature of the jth preliminary pixel and the importance weight of the undulation is used as the unevenness influence factor of the jth preliminary pixel. The unevenness influence factor of the jth preliminary pixel carefully reflects and quantifies the influence of the flatness of the glass surface on the edge.
[0072] It should be noted that the greater the regional non-flatness characteristics and the weight of the undulation importance of the prepared pixel point, the greater the possibility that the edge of the glass to be coated with glue where the prepared pixel point is located has undulations, and thus the greater the impact on the coating effect, and the greater the non-flatness influence factor of the prepared pixel point. Among them, the acquisition flow chart of the non-flatness influence factor is as follows: Figure 4 shown.
[0073] S3: extracting edge features of the glue-coated glass according to the distribution of the prepared pixel points, and obtaining edge influence feature factors of each prepared pixel point in combination with the non-flatness influence factor.
[0074] In the process of glass gluing, the geometric shape and curvature of the boundary have a significant impact on the gluing effect. Gluing on straight boundaries is relatively simple, but attention should be paid to the slight unevenness of the glass surface. If there are surface unevenness near the path points, it will affect the uniformity of gluing. Attention should be paid to the flatness of the surface to avoid glue accumulation or shortage. Boundaries with complex geometric shapes, such as multi-angle or irregular shapes with small curvature boundaries, will make gluing path planning more complicated.
[0075] Based on the above analysis, corner point detection is performed on the target image to obtain all corner points in the target image; the shortest Euclidean distance between the j-th prepared pixel point and all corner points is calculated and recorded as the feature distance; it should be noted that if the j-th prepared pixel point itself is a corner point, the shortest Euclidean distance is 0.
[0076] In some embodiments of the present application, the Harris corner detection algorithm is selected, which is a well-known technology and will not be described in detail in the present application; in addition, the implementer may select other corner detection or feature point detection algorithms, and the present application does not limit this. Figure 5 shown.
[0077] A local window is preset with the jth prepared pixel point as the center, wherein the local window is increased by a preset step size and a preset number of increases each time, so as to facilitate the analysis of the features of the edge skeleton in different local windows, so that the edge features extracted subsequently are more accurate.
[0078] In some embodiments of the present application, the size of the local window of the preset pixel point is ; The preset step size is 2; The preset number of increases is 3, that is, the size of the maximum local window of the prepared pixel point is .
[0079] It should be noted that when obtaining a local window, if the prepared pixel points are close to the boundary, it may cause the absence of pixel points in some areas of the local window. At this time, only the pixel values of the pixels existing in the window are analyzed, and the pixel values of the areas where there are no pixel points are filled with 0.
[0080] For the local window of the jth prepared pixel point, the edge skeleton curvature after each growth of the local window is obtained, and the square value of the difference between the edge skeleton curvature of each growth of the local window and the last growth is recorded as the edge skeleton curvature change factor after each growth. The edge skeleton curvature change factor can measure the degree of change of the skeleton shape when the local window changes. The ratio of the number of prepared pixels after each growth of the local window to the total number of pixels in the local window is used as the edge skeleton density after each growth. The edge skeleton density can measure the proportion of the edge skeleton in the local window. If the proportion of the edge skeleton in the local window is greater, the possibility of the shape of the edge skeleton being complex is greater. Calculate the product of the edge skeleton density after each growth and its corresponding growth number, and use the mapping of the opposite number of the product under the exponential function with a natural constant as the base as the density weight; calculate the cumulative sum of the product of the density weight and the edge skeleton curvature change factor under all growth times of the local window as the skeleton complexity feature factor of the jth prepared pixel point.
[0081] The edge influence characteristic factor of the j-th preliminary pixel point has a consistent trend relationship with the skeleton complexity characteristic factor, the characteristic distance, and the non-flatness influence factor.
[0082] It should be noted that a consistent trend relationship is specifically manifested in that as one variable increases, another variable also increases.
[0083] Based on the consistent trend relationship, implementers can flexibly select appropriate methods to determine the variables, and this application does not impose any restrictions on this.
[0084] In some embodiments of the present application, the normalized result of the product of the skeleton complexity feature factor, the shortest Euclidean distance and the non-flatness impact factor of the j-th pixel point is used as the edge impact feature factor of the j-th pixel point. It further reflects the geometric complexity of the edge skeleton where the j-th preliminary pixel point is located, including the curvature change and edge density, and considers the relative position of the preliminary pixel point and the nearest unstable area, and quantifies the importance of the j-th preliminary pixel point on the edge path. Among them, the flow chart for obtaining the edge feature impact factor is as follows: Figure 6 shown.
[0085] S4: Screening all the prepared pixel points according to the edge influence characteristic factor, and combining with the optimization algorithm to obtain the final gluing path.
[0086] The edge influence characteristic factors of all the prepared pixels are obtained, the influence threshold is set, and the prepared pixels whose edge influence characteristic factors are greater than the influence threshold are used as path points during gluing.
[0087] In some embodiments of the present application, the impact threshold is set to 0.4, which is convenient for screening the preliminary pixel points and taking the preliminary pixel points with obvious edge features as path points.
[0088] A random search algorithm is used according to all path points to obtain the gluing path; in some embodiments of the present application, a simple simulated annealing algorithm is selected, and its initial temperature parameter is set to the area of the glass to be glued, and the termination temperature parameter is set to 0; the temperature decay rate is the average gluing speed of the gluing robot; the cost function is set to the ratio of the glued area to the gluing speed; and the motion constraint condition is set to no repeated gluing and no omission of gluing.
[0089] It should be noted that the simple simulation algorithm is a well-known technology and will not be described in detail in this application.
[0090] In addition, at each path point, since the glue coating robot needs to change direction to continue gluing, a buffer zone is set around each path point to ensure that the glue coating robot does not miss any gluing when turning; the selected path points are fine-tuned according to the width of the glue coating head and the gluing speed, such as using an arc as a transition between two straight line segments, adjusting the spacing between path points, and discarding path points that have been covered. When the glue coating robot changes direction after passing through a path point, the path point with the largest edge influence characteristic factor among all path points that meet the requirements of non-repetition and non-omission of glue coating will be used as the next path point to ensure uniformity and continuity of glue coating and achieve gluing of large-size glass.
[0091] The embodiment of the present application provides a method for gluing large-size glass, the method comprising: obtaining point cloud data of the glass to be glued, marking the contour point cloud data, determining the contour of the glue, and conducting subsequent analysis based on the contour features; mapping the point cloud data onto a two-dimensional plane, and performing visual analysis with the help of the two-dimensional nature of the image, and more easily extracting the features and information of the prepared pixel points; obtaining the unstable area on the glass during the gluing process, constructing a non-flatness influencing factor, reflecting the microscopic unevenness characteristics of the glass surface, thereby realizing accurate quantification and evaluation of the flatness of the glass surface; and addressing the geometric complexity of the edge of the glass to be glued. , construct edge influence characteristic factors to reflect the curvature changes and edge density characteristics of the edge skeleton, so as to achieve accurate evaluation and optimization of the geometric complexity of the gluing path; by calculating the edge influence characteristic factors of each prepared pixel point and setting thresholds for screening, the accuracy and efficiency of gluing path selection can be improved to ensure that only path points that have a significant impact on the gluing quality are selected; through path point selection based on edge influence characteristic factors, the optimal planning of the gluing path is achieved to ensure the uniformity and continuity of the gluing process, and the gluing of glass of different shapes and sizes can be completed and the gluing quality can be guaranteed, while reducing the waste of glue materials and improving production efficiency.
[0092] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.
[0093] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the basic features of the present application. Therefore, no matter from which point of view, the above embodiments of the present application should be regarded as exemplary and non-restrictive.
Claims
1. A method for coating large-size glass, characterized in that: The method comprises the following steps: Obtaining point cloud data of the glass to be coated with glue, marking the contour point cloud data; mapping the point cloud data to be coated with glue and the contour point cloud data on a two-dimensional plane as a target image and preliminary pixel points of the target image respectively; According to the local pixel value distribution characteristics of all pixels in the target image, the unstable area is obtained, and a preset window is obtained with the pixel as the center. The standard deviation of the corresponding level of the pixel in the preset window is recorded as the smoothness factor of the pixel, and all pixels with non-zero smoothness factors in the target image are recorded as undulating pixels; according to the distance characteristics and shape characteristics of each prepared pixel and the corresponding unstable area, the non-flatness influencing factor of each prepared pixel is obtained; the shortest Euclidean distance and the longest Euclidean distance of each prepared pixel and the center point of all unstable areas are obtained; the unstable area corresponding to the shortest Euclidean distance is taken as the unstable area corresponding to the prepared pixel; according to the distance characteristics of the prepared pixel and the corresponding unstable area, the unstable area corresponding to the prepared pixel is taken as the unstable area corresponding to the prepared pixel; The fluctuation importance weight of the prepared pixel is obtained by the ratio of the area of the unsteady area to the maximum area of the unsteady area and the ratio of the shortest Euclidean distance to the longest Euclidean distance of the prepared pixel; in the unsteady area corresponding to the prepared pixel, the non-flat feature factor of the fluctuation pixel is obtained according to the difference between the flatness factor of the fluctuation pixel and the mean of all flatness factors, and the ratio of the variance of all flatness factors is calculated; the sum of all non-flat feature factors in the unsteady area corresponding to the prepared pixel is taken as the regional non-flat feature of the prepared pixel; the product of the regional non-flat feature of each prepared pixel and the fluctuation importance weight is taken as the non-flatness influencing factor of each prepared pixel; According to the distribution of the prepared pixel points, the edge features of the coated glass are extracted, and the edge influence feature factor of each prepared pixel point is obtained in combination with the non-flatness influencing factor; the shortest distance between the prepared pixel point and all corner points is used as the feature distance of the prepared pixel point; a local window is preset with the prepared pixel point as the center; the edge skeleton curvature change factor is obtained according to the difference between the edge skeleton curvature of each growth of the local window and the last growth; the density weight is obtained according to the ratio of the number of prepared pixel points to the number of pixels in the local window after each growth of the local window and the number of growth times; the skeleton complexity feature factor is obtained according to the density weight and the edge skeleton curvature change factor under all growth times; the edge influence feature factor of the prepared pixel point is obtained according to the feature distance of the prepared pixel point, the skeleton complexity feature factor, and the non-flatness influencing factor; All the prepared pixel points are screened according to the edge influence characteristic factors, and the final gluing path is obtained by combining the optimization algorithm.
2. A method for coating glue for large-size glass as claimed in claim 1, characterized in that: The step of obtaining the unstable region according to the local pixel value distribution characteristics of all pixels in the target image comprises: Divide the pixel values of all pixels in the target image into a preset number of levels; Calculate the standard deviation of the corresponding levels of all pixels in the preset window and record it as the smoothness factor of each pixel; record all pixels in the target image whose smoothness factor is not zero as undulating pixels, and use them as nodes to obtain a connection graph using a priority search algorithm; record each area in the connection graph as an unstable area.
3. A method for coating glue for large-size glass as claimed in claim 2, characterized in that: The non-flatness influencing factor of each prepared pixel point is obtained according to the distance characteristics and shape characteristics of each prepared pixel point and the unstable area, specifically: Obtain the shortest Euclidean distance and the longest Euclidean distance between each prepared pixel and the center point of all unstable areas; and use the unstable area corresponding to the shortest Euclidean distance of each prepared pixel as the unstable area corresponding to each prepared pixel; According to the distance distribution and area distribution of each pixel and its corresponding unstable area, the fluctuation importance weight of each prepared pixel is obtained; According to the distribution of the undulating pixels in the unstable area corresponding to each prepared pixel point, the regional non-smoothness feature of each prepared pixel point is obtained; The product of the regional unevenness feature of each prepared pixel point and the undulation importance weight is used as the unevenness influencing factor of each prepared pixel point.
4. A method for coating glue for large-size glass as claimed in claim 3, characterized in that: The step of obtaining the fluctuation importance weight of each prepared pixel point is specifically as follows: The ratio of the area of the unstable region corresponding to each prepared pixel point to the maximum area of all unstable regions is used as the fluctuation regional factor of each prepared pixel point; The ratio of the shortest Euclidean distance to the longest Euclidean distance of each prepared pixel point is used as the fluctuation distance factor of each prepared pixel point; Obtaining the ups and downs importance weight of each prepared pixel point according to the ups and downs regional factor and the ups and downs distance factor of each prepared pixel point; The fluctuation importance weight of each prepared pixel point is in increasing relationship with the fluctuation regional factor and the fluctuation distance factor.
5. A method for coating glue for large-size glass as claimed in claim 3, characterized in that: The obtaining of the regional non-flat features of each prepared pixel point is specifically as follows: For the non-stationary area corresponding to each prepared pixel point, the average value and variance of the smoothness factors of all the fluctuating pixels are obtained; the square value of the difference between the smoothness factor of each fluctuating pixel point and the average value is calculated, the inverse of the ratio of the square value of the difference to the variance is calculated, and the mapping of the inverse under the exponential function is recorded as the non-smooth characteristic factor of each fluctuating pixel point; The cumulative sum of all the non-smooth feature factors in the non-smooth region corresponding to each prepared pixel point is obtained, and recorded as the regional non-smooth feature of each prepared pixel point.
6. A method for coating glue for large-size glass as claimed in claim 1, characterized in that: The step of obtaining the edge influence characteristic factor of each prepared pixel point comprises the following steps: Corner point detection is performed on the target image, and the characteristic distance of each prepared pixel point is obtained according to the distance distribution between each prepared pixel point and the corner point; The local window is increased each time with a preset step size and a preset number of increases; according to the change of the local window of each prepared pixel point and the change of the shape feature of the edge, the skeleton complex feature factor of each prepared pixel point is obtained; Obtaining an edge influence characteristic factor of each prepared pixel point according to the characteristic distance of each prepared pixel point, the skeleton complexity characteristic factor, and the non-flatness influence factor; The edge influence characteristic factor of each prepared pixel point is in a consistent trend relationship with the skeleton complexity characteristic factor, the characteristic distance and the non-flatness influence factor.
7. A method for coating glue for large-size glass as claimed in claim 6, characterized in that: The method of obtaining the characteristic distance of each prepared pixel point is specifically as follows: Corner point detection is performed on the target image to obtain all corner points in the target image, and the shortest Euclidean distance between each prepared pixel point and all corner points is used as the feature distance of each prepared pixel point.
8. A method for coating glue for large-size glass as claimed in claim 6, characterized in that: The specific steps of obtaining the complex characteristic factor of the skeleton of each prepared pixel point include: For the local window of each prepared pixel point, the edge skeleton curvature after each growth of the local window is obtained, and the square value of the difference between the edge skeleton curvature of each growth of the local window and the last growth is recorded as the edge skeleton curvature change factor after each growth; The density weight after each growth is obtained according to the edge skeleton curvature change factor after each growth and the number of prepared pixels in the corresponding local window; The cumulative sum of the product of the density weight and the edge skeleton curvature change factor under all growth times of the local window is calculated as the skeleton complex feature factor of each prepared pixel point.
9. A method for coating glue for large-size glass as claimed in claim 8, characterized in that: The density weight after each increase is obtained as follows: The ratio of the number of prepared pixels after each growth of the local window to the total number of pixels in the local window is used as the edge skeleton density after each growth; The product of the edge skeleton density after each growth and the corresponding growth times is calculated, and the mapping of the inverse of the product under the exponential function is used as the density weight.
10. The method for coating glue for large-size glass according to claim 1, characterized in that: The final gluing path is obtained as follows: The prepared pixel points whose edge influence characteristic factors are greater than the preset influence threshold are used as path points during gluing, and the gluing path is obtained by using a random search algorithm based on all the path points.
