Automobile covering part spraying track planning method and system based on machine vision

Abstract

The invention relates to the technical field of machine vision, and particularly discloses an automobile covering part spraying track planning method and system based on machine vision, and the method comprises the steps: carrying out the denoising and enhancement processing of complete spraying surface images of all parts of covering components in an automobile covering part, and obtaining the effective spraying surface images of all parts of covering components; performing contour recognition on the effective spraying surface image of each partial covering component to obtain an effective spraying surface contour of each partial covering component; determining a planned spraying track of each partial covering component based on the effective spraying surface contour of each partial covering component; the method is used for denoising and enhancing complete spraying surface images of all partial covering components in an automobile covering part to obtain effective spraying surface images of all partial covering components. And based on the effective spraying surface contours in the effective spraying surface images of all the partial covering components, reasonable planning of spraying tracks of all the partial covering components in the automobile covering part is realized.

Description

Technical Field

[0001] The invention relates to the technical field of machine vision, and in particular to a method and system for planning the spraying trajectory of automobile covering parts based on machine vision. Background Art

[0002] At present, automobile covering parts refer to parts such as the shell, engine cover, wheel hub, fender, etc. installed on the car. The spray trajectory planning method of automobile covering parts refers to the method of trajectory planning and controlling the spray process of automobile covering parts through computer algorithms and other technologies. The spray quality of automobile covering parts is directly related to the quality and appearance of the car. Therefore, automobile manufacturers need to adopt advanced spraying technology to spray covering parts with high quality. The spray trajectory planning method of automobile covering parts can help automobile manufacturers better control the spraying process and ensure the spraying quality. In the spraying process, the uniformity of the coating is not only closely related to the characteristics of the coating itself, the spraying model of the spraying robot, etc., but also to factors such as the spraying process parameters and the spraying operation trajectory. Some technologies for the spray trajectory planning method of automobile covering parts have been proposed. For example, a spraying robot based on machine vision and high-precision sensing technology can be used to complete the spraying of automobile covering parts. Automobile covering parts are products with large-area free-form surfaces, and they have strict requirements on the uniformity of the coating. Therefore, the optimization technology of spraying trajectory based on machine vision has become one of the key technologies of automatic spraying robots.

[0003] However, when the camera is shooting in a factory environment, a variety of factors may cause noise in the captured image. Among them, environmental factors are the most common reason. Light, high temperature, humidity, etc. may affect the quality of the photo, causing the image to produce Gaussian noise, salt and pepper noise and other noise. In addition, when the camera's photosensitive element is exposed to a high temperature environment, it will also cause dark current noise in the image. In the spray trajectory technology based on machine vision, various noises in the cover image will seriously affect the accuracy and operability of the subsequent planning of the cover spray trajectory.

[0004] Therefore, the present invention proposes a method and system for planning the spraying trajectory of automobile covering parts based on machine vision. Summary of the invention

[0005] The present invention provides a method and system for planning the spraying trajectory of an automobile covering part based on machine vision, which is used to obtain effective spraying surface images of all partial covering components in the automobile covering part by denoising and enhancing the complete spraying surface images of all partial covering components in the automobile covering part, thereby ensuring the accuracy and operability of the spraying trajectory planning of the covering part, and realizing reasonable planning of the spraying trajectory of all partial covering components in the automobile covering part based on the effective spraying surface contours in the effective spraying surface images of all partial covering components, thereby ensuring the spraying quality of the automobile covering part.

[0006] The present invention provides a method for planning the spraying trajectory of automobile panels based on machine vision, comprising:

[0007] S1: De-noising and enhancing the complete spray surface images of all partially covered components of the automobile covering parts to obtain effective spray surface images of all partially covered components;

[0008] S2: performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component;

[0009] S3: Determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

[0010] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S1: denoising and enhancing the complete spraying surface images of all partial covering components in the automobile covering parts to obtain the effective spraying surface images of all partial covering components, including:

[0011] S101: Acquire complete spray surface images of all partial covering components in the automobile covering part;

[0012] S102: performing multiple hypothetical grayscale processing on the complete spray surface image of each partially covered component based on a weighted average method and screening the processing results to obtain all valid hypothetical spray surface grayscale images of each partially covered component;

[0013] S103: performing multiple assumption filtering processes on all valid assumption spray surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spray surface filtering images of each partially covered component;

[0014] S104: performing multiple hypothetical binarization processing, white spot and burr elimination, and processing result screening on all valid hypothetical spray surface filter images of each partially covered component to obtain a valid spray surface image of each partially covered component.

[0015] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S102: performing multiple hypothetical grayscale processing on the complete spraying surface image of each partially covered component and screening the processing results based on the weighted average method to obtain all valid hypothetical spraying surface grayscale images of each partially covered component, including:

[0016] Based on the constraint that the sum of the weight values ​​corresponding to the three channel values ​​is 1, a preset number of groups of hypothetical weight values ​​are generated;

[0017] The sum of the products of the three channel values ​​of each pixel point in the complete spray surface image of each partially covered component and the corresponding preset weight value in each set of preset weight values ​​is regarded as the grayscale value of each pixel point in the complete spray surface image of each partially covered component relative to the corresponding set of assumed weight values;

[0018] Generate a hypothetical grayscale image of each partially covered component relative to each set of hypothetical weight values ​​based on the grayscale value of each pixel in the complete spray surface image of each partially covered component relative to each set of hypothetical weight values;

[0019] Among all the hypothetical grayscale images of each partially covered component, all valid hypothetical spray surface grayscale images of each partially covered component are screened out.

[0020] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision selects all valid hypothetical spraying surface grayscale images of each partially covered component from all hypothetical grayscale images of each partially covered component, including:

[0021] The average of the grayscale values ​​of all pixels in each hypothetical grayscale image of each partially covered component is regarded as the overall brightness of each hypothetical grayscale image;

[0022] All the hypothetical grayscale images of each partially covered component whose overall brightness is not less than the too dark threshold and not greater than the too bright threshold are regarded as all the valid hypothetical spray surface grayscale images of each partially covered component.

[0023] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S103: performing multiple assumption filtering processes on all valid assumption spraying surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spraying surface filter images of each partially covered component, including:

[0024] Performing linear mean filtering on all valid hypothetical spray surface grayscale images of each partially covered component to obtain a first hypothetical filtered image of each valid hypothetical spray surface grayscale image of each partially covered component;

[0025] Performing median filtering on all valid hypothetical spray-painted surface grayscale images of each partially covered component to obtain a second hypothetical filtered image of each valid hypothetical spray-painted surface grayscale image of each partially covered component;

[0026] Inputting the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image into the image clarity evaluation model respectively, obtaining the clarity evaluation value of the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image;

[0027] Inputting the first assumption filtered image and the second assumption filtered image of each valid assumption spray surface grayscale image into the edge information integrity evaluation model respectively, obtaining the edge information integrity evaluation value of the first assumption filtered image and the edge information integrity evaluation value of the second assumption filtered image of each valid assumption spray surface grayscale image;

[0028] Based on the clarity evaluation value and edge information integrity evaluation value of the first hypothetical filtered image of all valid hypothetical spray surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second hypothetical filtered image, all valid hypothetical spray surface filter images of each partially covered component are obtained.

[0029] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision obtains all valid hypothetical spraying surface filter images of each partially covered component based on the clarity evaluation value and edge information integrity evaluation value of the first hypothetical filtered image of all valid hypothetical spraying surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second hypothetical filtered image, including:

[0030] The first hypothetical filtered image or the second hypothetical filtered image having the largest weighted sum of the clarity evaluation value and the edge information integrity evaluation value among the first hypothetical filtered image and the second hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component is regarded as the third hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component;

[0031] All third hypothetical filtered images of each partially covered component, whose weighted sum of clarity evaluation value and edge information integrity evaluation value is not less than a threshold, are regarded as all valid hypothetical spray surface filter images of each partially covered component.

[0032] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S104: performing multiple hypothetical binarization processing, white spot and burr elimination and processing result screening on all valid hypothetical spraying surface filter images of each partially covered component to obtain the valid spraying surface image of each partially covered component, including:

[0033] Based on the Otsu algorithm, all valid hypothetical spray surface filter images of each partially covered component are binarized to obtain all hypothetical binary images of each partially covered component;

[0034] Determine the region composed of all pure black pixels in each assumed binary image as the first region of each assumed binary image, and at the same time, determine the region composed of all pure white pixels in each assumed binary image as the second region of each assumed binary image;

[0035] The first area or the second area with the largest number of pixels in each assumed binary image is regarded as a foreground area, and the first area or the second area with the least number of pixels in each assumed binary image is regarded as a background area;

[0036] The ratio of the total number of pixels in the background area to the total number of pixels in the foreground area in each hypothetical binary image is regarded as the validity of each hypothetical binary image, and all hypothetical binary images whose validity is not less than the validity threshold among all hypothetical binary images of each partially covered component are regarded as all valid binary images of each partially covered component;

[0037] All valid binary images of each partially covered component are subjected to white spot and burr elimination and processing result screening to obtain a valid spray surface image of each partially covered component.

[0038] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision performs white spot and burr elimination and processing result screening on all valid binary images of each partially covered component to obtain the valid spraying surface image of each partially covered component, including:

[0039] Determine a reasonable structure element size based on the maximum single-side size of the foreground area of ​​each valid binary image of each partially covered component;

[0040] Based on the reasonable structure element size, each effective binary image of each partially covered component is opened until each effective binary image becomes a single connected region image, and a spraying region image corresponding to each effective binary image is obtained;

[0041] The overlapping area of ​​the spraying area images corresponding to all valid binary images of each partially covered component is regarded as the necessary area of ​​each partially covered component, and all pixel points except the necessary area in each spraying area image of the corresponding partially covered component are regarded as all accidental pixel points of the corresponding spraying area image;

[0042] In all spraying area images of each partially covered component, the occasional pixel points whose appearance frequency ratio is not less than the preset frequency ratio threshold are regarded as the inevitable pixel points of the corresponding partially covered component, and the inevitable area and all inevitable pixel points of each partially covered component are summarized to obtain the inevitable spraying area of ​​each partially covered component;

[0043] The necessary spraying area of ​​each partially covered component is expanded until a single connected area with a smooth contour is obtained, and the obtained single connected area with a smooth contour is regarded as the effective spraying surface image of each partially covered component.

[0044] Preferably, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S3: determining the planned spraying trajectory of each partial covering component based on the effective spraying surface profile of each partial covering component, comprising:

[0045] S301: determining an effective spraying area and an effective spraying surface contour shape of each partially covered component based on an effective spraying surface contour of each partially covered component;

[0046] S302: Determine a planned spraying trajectory of each partially covered component based on the effective spraying area and the effective spraying surface contour shape and the spraying range radius of each partially covered component.

[0047] The present invention provides a system for planning the spraying trajectory of an automobile panel based on machine vision, which is used to execute the method for planning the spraying trajectory of an automobile panel based on machine vision described in any one of Embodiments 1 to 9, comprising:

[0048] An image processing module is used to perform denoising and enhancement processing on the complete spray surface images of all partially covered components in the automobile covering parts to obtain effective spray surface images of all partially covered components;

[0049] A contour determination module, used for performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component;

[0050] The trajectory planning module is used to determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

[0051] The beneficial effects of the present invention compared with the prior art are as follows: by denoising and enhancing the complete spray surface images of all partial covering components in the automobile covering, the effective spray surface images of all partial covering components are obtained, which ensures the accuracy and operability of the spray trajectory planning of the covering components, and based on the effective spray surface contours in the effective spray surface images of all partial covering components, the spray trajectory of all partial covering components in the automobile covering is reasonably planned, thereby ensuring the spraying quality of the automobile covering.

[0052] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0053] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 This is a flow chart of a method for planning the spraying trajectory of automobile panels based on machine vision in an embodiment of the present invention;

[0056] Figure 2 It is a schematic diagram of the spraying steps of a partial covering component in an embodiment of the present invention;

[0057] Figure 3 is a grayscale image of an effective hypothetical sprayed surface of a partially covered component in an embodiment of the present invention;

[0058] Figure 4 An example diagram of a valid binary image of a partially covered component in an embodiment of the present invention;

[0059] Figure 5 An example diagram of an effective spray surface image of a partially covered component in an embodiment of the present invention;

[0060] Figure 6 Schematic diagram of the spraying trajectory in an embodiment of the present invention;

[0061] Figure 7 Schematic diagram of the internal functional modules of the machine vision-based automobile panel painting trajectory planning system in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0063] Embodiment 1:

[0064] The present invention provides a method for planning the spraying trajectory of automobile panels based on machine vision, referring to Figure 1 and 2 ,include:

[0065] S1: De-noising and enhancing the complete spray surface images of all partially covered components of the automobile covering parts to obtain effective spray surface images of all partially covered components;

[0066] S2: performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component;

[0067] S3: Determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

[0068] In this embodiment, the automobile covering part is composed of a plurality of partial covering components, wherein the partial covering components include, for example, a housing, an engine cover, a wheel hub, a fender and other parts.

[0069] In this embodiment, the complete spraying surface image is an image including the complete structure of the side of the partial covering component that needs to be sprayed.

[0070] In this embodiment, denoising includes multiple assumptions of grayscale processing and processing result screening, multiple assumptions of filtering processing and processing result screening.

[0071] In this embodiment, the enhancement process includes multiple hypothetical binarization processes, white spot and burr elimination, and processing result screening.

[0072] In this embodiment, the effective spraying surface image is an image including all actual component areas that partially cover the component and ultimately need to be sprayed.

[0073] In this embodiment, the Canny edge detection operator is used to perform contour recognition on the effective spray surface image of each partially covered component.

[0074] In this embodiment, the effective spray surface contour is the outer contour line of the effective spray surface image.

[0075] In this embodiment, the planned spraying trajectory of the partially covering component is the spraying trajectory finally planned for the partially covering component, wherein:

[0076] refer to Figure 2 The spraying process is divided into two parts. The first part is the spraying process of "partially covering the edge of the component to be sprayed and part of the plate surface", and the second part is the spraying process of partially covering the central plate surface of the component to be sprayed.

[0077] The beneficial effects of the above technology are: by denoising and enhancing the complete spray surface images of all partially covering components in the automobile covering, the effective spray surface images of all partially covering components are obtained, and based on the effective spray surface contours in the effective spray surface images of all partially covering components, the reasonable planning of the spray trajectory of all partially covering components in the automobile covering is achieved.

[0078] Embodiment 2:

[0079] Based on Example 1, a method for planning the spraying trajectory of automobile covering parts based on machine vision is provided. S1: De-noising and enhancing the complete spraying surface images of all partial covering components in the automobile covering parts are performed to obtain the effective spraying surface images of all partial covering components, referring to Figure 3 and 5 ,include:

[0080] S101: Acquire complete spray surface images of all partial covering components in the automobile covering part;

[0081] S102: performing multiple hypothetical grayscale processing on the complete spray surface image of each partially covered component based on a weighted average method and screening the processing results to obtain all valid hypothetical spray surface grayscale images of each partially covered component;

[0082] S103: performing multiple assumption filtering processes on all valid assumption spray surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spray surface filtering images of each partially covered component;

[0083] S104: performing multiple hypothetical binarization processing, white spot and burr elimination, and processing result screening on all valid hypothetical spray surface filter images of each partially covered component to obtain a valid spray surface image of each partially covered component.

[0084] In this embodiment, the multiple times of assumed gray-scaling processing and the processing result screening in the processing result screening is to screen the results obtained after the multiple times of assumed gray-scaling processing are performed.

[0085] In this embodiment, the multiple assumption filtering processes and the processing result screening in the processing result screening are to screen the results obtained after the multiple assumption filtering processes are performed.

[0086] In this embodiment, the white point is a white tiny noise area or noise point in the effective hypothetical spray surface filter image, and generally belongs to the background area in the effective hypothetical spray surface filter image.

[0087] In this embodiment, the burr is a black noise area or noise point on the outline of the black area in the effective hypothetical spray surface filter image, and generally belongs to the foreground area of ​​the effective hypothetical spray surface filter image.

[0088] In this embodiment, the processing result screening in performing multiple times of hypothetical binarization processing, eliminating white spots and burrs, and screening the processing results is to screen the results obtained after performing multiple times of hypothetical binarization processing.

[0089] The beneficial effects of the above technology are: by performing multiple hypothetical grayscale processing and processing result screening, multiple hypothetical filtering processing and processing result screening, multiple hypothetical binarization processing, white spot and burr elimination and processing result screening on the complete spray surface image of the partial covering components in the automobile covering parts, effective denoising and enhancement processing of the complete spray surface image of all partial covering components in the automobile covering parts is achieved, and the rationality of the planning of the subsequent spray trajectory of the partial covering components is also guaranteed.

[0090] Embodiment 3:

[0091] Based on Example 2, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S102: performing multiple hypothetical grayscale processing on the complete spraying surface image of each partially covered component and screening the processing results based on the weighted average method, to obtain all valid hypothetical spraying surface grayscale images of each partially covered component, including:

[0092] Based on the constraint that the sum of the weight values ​​corresponding to the three channel values ​​is 1, a preset number of groups of hypothetical weight values ​​are generated;

[0093] The sum of the products of the three channel values ​​of each pixel point in the complete spray surface image of each partially covered component and the corresponding preset weight value in each set of preset weight values ​​is regarded as the grayscale value of each pixel point in the complete spray surface image of each partially covered component relative to the corresponding set of assumed weight values;

[0094] Generate a hypothetical grayscale image of each partially covered component relative to each set of hypothetical weight values ​​based on the grayscale value of each pixel in the complete spray surface image of each partially covered component relative to each set of hypothetical weight values;

[0095] Among all the hypothetical grayscale images of each partially covered component, all valid hypothetical spray surface grayscale images of each partially covered component are screened out.

[0096] In this embodiment, based on the constraint that the sum of the weight values ​​corresponding to the three channel values ​​is 1, a preset number of groups of hypothetical weight values ​​are generated, that is, the sum of the weight values ​​corresponding to the three channel values ​​contained in each group of hypothetical weight values ​​generated is 1, and the preset number can be 100, for example.

[0097] In this embodiment, the three channel values ​​are R value, G value, and B value.

[0098] In this embodiment, based on the grayscale value of each pixel in the complete spray surface image of each partially covered component relative to each set of assumed weight values, a hypothetical grayscale image of each partially covered component relative to each set of assumed weight values ​​is generated, namely:

[0099] The pixel value of each pixel in the complete spray surface image of each partially covered component is set to the grayscale value of the corresponding pixel in the complete spray surface image of the corresponding partially covered component relative to a single set of assumed weight values, to obtain a hypothetical grayscale image of each partially covered component relative to each set of assumed weight values.

[0100] The beneficial effect of the above technology is: based on the generated multiple sets of hypothetical weight values, multiple hypothetical grayscale processing is performed on the complete spray surface image of each partially covered component through the weighted average method to ensure the image effect of the final effective hypothetical spray surface grayscale image.

[0101] Embodiment 4:

[0102] On the basis of Example 3, the method for planning the spraying trajectory of automobile covering parts based on machine vision screens out all valid hypothetical spraying surface grayscale images of each partially covered component from all hypothetical grayscale images of each partially covered component, including:

[0103] The average of the grayscale values ​​of all pixels in each hypothetical grayscale image of each partially covered component is regarded as the overall brightness of each hypothetical grayscale image;

[0104] All the hypothetical grayscale images of each partially covered component whose overall brightness is not less than the too dark threshold and not greater than the too bright threshold are regarded as all the valid hypothetical spray surface grayscale images of each partially covered component.

[0105] In this embodiment, the overall brightness and darkness of the assumed grayscale image is a numerical value representing the overall brightness and darkness of the assumed grayscale image. Because a grayscale image that is too bright or too dark as a whole will cause distortion of the grayscale image, it is necessary to use this numerical value to screen the numerous assumed grayscale images obtained to ensure the display effect of the final effective assumed spray surface grayscale image.

[0106] In this embodiment, the over-darkness threshold is a preset effective assumed spray surface grayscale image overall brightness and darkness value that cannot be less than.

[0107] In this embodiment, the overbrightness threshold is a preset effective assumed value that the overall brightness of the grayscale image of the sprayed surface cannot be greater than.

[0108] The beneficial effect of the above technology is: by obtaining the overall brightness and darkness of each hypothetical grayscale image, and screening the hypothetical grayscale images with the overall brightness and darkness not less than the too dark threshold and not greater than the too bright threshold as the constraint, the excessive distortion of the final effective hypothetical spray surface grayscale image is avoided, that is, the display effect of the final effective hypothetical spray surface grayscale image is guaranteed.

[0109] Embodiment 5:

[0110] On the basis of Example 2, the method for planning the spraying trajectory of automobile covering parts based on machine vision, S103: performing multiple assumption filtering processes on all valid assumption spraying surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spraying surface filtering images of each partially covered component, including:

[0111] Performing linear mean filtering on all valid hypothetical spray surface grayscale images of each partially covered component to obtain a first hypothetical filtered image of each valid hypothetical spray surface grayscale image of each partially covered component;

[0112] Performing median filtering on all valid hypothetical spray-painted surface grayscale images of each partially covered component to obtain a second hypothetical filtered image of each valid hypothetical spray-painted surface grayscale image of each partially covered component;

[0113] Inputting the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image into the image clarity evaluation model respectively, obtaining the clarity evaluation value of the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image;

[0114] Inputting the first assumption filtered image and the second assumption filtered image of each valid assumption spray surface grayscale image into the edge information integrity evaluation model respectively, obtaining the edge information integrity evaluation value of the first assumption filtered image and the edge information integrity evaluation value of the second assumption filtered image of each valid assumption spray surface grayscale image;

[0115] Based on the clarity evaluation value and edge information integrity evaluation value of the first hypothetical filtered image of all valid hypothetical spray surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second hypothetical filtered image, all valid hypothetical spray surface filter images of each partially covered component are obtained.

[0116] In this embodiment, linear mean filtering is performed on all valid hypothetical spray surface grayscale images of each partially covered component to obtain a first hypothetical filtered image of each valid hypothetical spray surface grayscale image of each partially covered component, which is:

[0117] The basic idea of ​​linear mean filtering is to establish a sliding window of a self-defined size with the target pixel as the center and move it on the image. The range outside the window except the target pixel is the neighborhood. The RGB values ​​of all pixels in the neighborhood are extracted for weighted average, and the average value is used as the new value of the target pixel. The matrix description is:

[0118]

[0119] This method determines the neighborhood range of the target pixel point, replaces the target pixel value with the RGB average value of all pixels in the neighborhood range, and completes the denoising of the black and white grayscale image of the fender. Assuming that the number of pixels of the grayscale image f(a,b) of the sprayed surface is a×b, the image after linear mean filtering and denoising is g(m,n), which can be expressed as:

[0120]

[0121] Where s is the coordinate set within the neighborhood, and M is the total number of pixels within the neighborhood.

[0122] In this embodiment, median filtering is performed on all valid hypothetical spray surface grayscale images of each partially covered component to obtain a second hypothetical filtered image of each valid hypothetical spray surface grayscale image of each partially covered component, including:

[0123] The principle of median filtering is to select the target pixel in the grayscale image of the effective hypothetical spray surface, sort the RGB values ​​of the odd number of pixels within a certain range around it, and then use the sorted middle value as the new value of the target pixel. This is helpful to prevent the image from being distorted, distorted, and having poor quality after filtering. The expression formula is:

[0124]

[0125] In the formula, g(i,j) represents the gray value of the pixel after median filtering, and x(i,j) represents the neighborhood I 2 All pixel values ​​within the range, Med means that the neighborhood I 2 Sort all pixel values ​​within the range and take the median value;

[0126] Compared with linear mean filtering, median filtering can better preserve image edge and detail information, and has a better suppression effect on common non-Gaussian noise such as salt and pepper noise.

[0127] In this embodiment, the image definition evaluation model is a pre-trained model that can output a definition evaluation value of “an image input to the model”;

[0128] The model is obtained by training a neural network model using a large number of pre-prepared images and their clarity evaluation values ​​calculated based on existing technologies as model input and model output respectively.

[0129] In this embodiment, the clarity evaluation value of the first hypothetical filtered image or the second hypothetical filtered image is a numerical value used to represent the degree of clarity of the first hypothetical filtered image or the second hypothetical filtered image.

[0130] In this embodiment, the edge information integrity evaluation model is a pre-trained model that can output an edge information integrity evaluation value of “an image input to the model”;

[0131] The model is obtained by training a neural network model using a large number of pre-prepared images and their edge information integrity calculated based on existing technologies as model input and model output respectively.

[0132] In this embodiment, the edge information completeness evaluation value of the first assumed filtered image or the second assumed filtered image is a numerical value used to represent the edge information completeness of the first assumed filtered image or the second assumed filtered image. The edge information includes, for example, the specific degree of change in the edge direction and the clarity of the edge pixels.

[0133] The beneficial effects of the above technology are as follows: by performing linear mean filtering and median filtering on all valid hypothetical spray surface grayscale images of each partially covered component respectively, multiple hypothetical filtering processing of all valid hypothetical spray surface grayscale images of each partially covered component is completed, and further, based on the clarity evaluation value and edge information integrity of the filtered image output by the image clarity evaluation model and the edge information integrity evaluation model as constraints, further effective screening of the numerous filtered images obtained after filtering processing is achieved, and then, after ensuring the image display effect of the obtained effective hypothetical spray surface filter image, the number of effective hypothetical spray surface filter images of each partially covered component is simplified.

[0134] Embodiment 6:

[0135] On the basis of Example 5, a method for planning the spraying trajectory of an automobile covering part based on machine vision is provided, based on the clarity evaluation value and edge information integrity evaluation value of the first assumed filtered image of all valid assumed spraying surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second assumed filtered image, to obtain all valid assumed spraying surface filter images of each partially covered component, including:

[0136] The first hypothetical filtered image or the second hypothetical filtered image having the largest weighted sum of the clarity evaluation value and the edge information integrity evaluation value among the first hypothetical filtered image and the second hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component is regarded as the third hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component;

[0137] All third hypothetical filtered images of each partially covered component, whose weighted sum of clarity evaluation value and edge information integrity evaluation value is not less than a threshold, are regarded as all valid hypothetical spray surface filter images of each partially covered component.

[0138] In this embodiment, the weighted sum of the clarity evaluation value and the edge information integrity evaluation value is:

[0139] The product of the clarity evaluation value and the corresponding weight value, and the product of the edge information integrity evaluation value and the corresponding weight value, are taken as their weighted sum value. The specific values ​​of the weight values ​​can be set according to the big data analysis results. For example, the weight values ​​of the clarity evaluation value and the edge information integrity evaluation value can be set to 0.5.

[0140] In addition, the sum of the weight value corresponding to the clarity evaluation value and the weight value corresponding to the edge information integrity evaluation value is set to 1, and the weight values ​​corresponding to the clarity evaluation values ​​of all first assumed filtered images and second assumed filtered images are the same, and the weight values ​​corresponding to the edge information integrity evaluation values ​​are also the same.

[0141] In this embodiment, the threshold is a value that the weighted sum of the clarity evaluation value and the edge information integrity evaluation value of the preset effective hypothetical spray surface filter image cannot be less than.

[0142] The beneficial effect of the above technology is: the weighted sum of the clarity evaluation value and the edge information completeness evaluation value of the first assumed filtered image and the second assumed filtered image of each effective assumed spray surface grayscale image of each partially covered component is used as a constraint condition, thereby ensuring that the image display effect of the effective assumed spray surface filter image finally screened out has good performance in terms of image clarity and edge information completeness.

[0143] Embodiment 7:

[0144] Based on Example 2, a method for planning the spraying trajectory of automobile covering parts based on machine vision, S104: performing multiple hypothetical binarization processing, white spot and burr elimination and processing result screening on all valid hypothetical spraying surface filter images of each partially covered component, to obtain the valid spraying surface image of each partially covered component, refer to Figure 4 ,include:

[0145] Based on the Otsu algorithm, all valid hypothetical spray surface filter images of each partially covered component are binarized to obtain all hypothetical binary images of each partially covered component;

[0146] Determine the region composed of all pure black pixels in each assumed binary image as the first region of each assumed binary image, and at the same time, determine the region composed of all pure white pixels in each assumed binary image as the second region of each assumed binary image;

[0147] The first area or the second area with the largest number of pixels in each assumed binary image is regarded as a foreground area, and the first area or the second area with the least number of pixels in each assumed binary image is regarded as a background area;

[0148] The ratio of the total number of pixels in the background area to the total number of pixels in the foreground area in each hypothetical binary image is regarded as the validity of each hypothetical binary image, and all hypothetical binary images whose validity is not less than the validity threshold in all hypothetical binary images of each partially covered component are regarded as all valid binary images of each partially covered component (e.g. Figure 4 ,Depend on Figure 4 It can be seen that the number of pixels in the foreground area is much larger than that in the background area, and its contour contains many burrs, while the foreground area contains many white spots).

[0149] All effective binary images of each partially covered component are subjected to white spot and burr elimination and processing result screening to obtain an effective spray surface image of each partially covered component.

[0150] In this embodiment, the effectiveness of the assumed binary image is a numerical value indicating the quality of the enhancement effect after the foreground image of the assumed binary image is subjected to image enhancement (ie, binarization processing).

[0151] In this embodiment, the validity threshold is a value that the validity of the valid binary image cannot be less than.

[0152] The beneficial effects of the above technology are: it realizes the binarization processing of all valid hypothetical spray surface filter images of each partially covered component, and obtains the quantitative parameter of effectiveness by statistically analyzing and comparing the number of pixels contained in the foreground area and background area of ​​all hypothetical binary images obtained after the binarization processing, thereby realizing the accurate quantification of the excellence of the enhancement effect after the foreground image of the hypothetical binary image is enhanced (i.e., binarization processing), thereby ensuring the effectiveness of the effective binary image finally obtained, and reducing the number of images that need to be subsequently eliminated of white spots and burrs and screened for processing results through the screening step.

[0153] Embodiment 8:

[0154] On the basis of Example 7, a method for planning the spraying trajectory of automobile covering parts based on machine vision is used to eliminate white spots and burrs and screen the processing results of all valid binary images of each partially covered component to obtain the effective spraying surface image of each partially covered component, including:

[0155] Determine a reasonable structure element size based on the maximum single-side size of the foreground area of ​​each valid binary image of each partially covered component;

[0156] Based on the reasonable structure element size, each effective binary image of each partially covered component is opened until each effective binary image becomes a single connected region image, and a spraying region image corresponding to each effective binary image is obtained;

[0157] The overlapping area of ​​the spraying area images corresponding to all valid binary images of each partially covered component is regarded as the necessary area of ​​each partially covered component, and all pixel points except the necessary area in each spraying area image of the corresponding partially covered component are regarded as all accidental pixel points of the corresponding spraying area image;

[0158] In all spraying area images of each partially covered component, the occasional pixel points whose appearance frequency ratio is not less than the preset frequency ratio threshold are regarded as the inevitable pixel points of the corresponding partially covered component, and the inevitable area and all inevitable pixel points of each partially covered component are summarized to obtain the inevitable spraying area of ​​each partially covered component;

[0159] The necessary spraying area of ​​each partially covered component is expanded until a single connected area with a smooth contour is obtained, and the obtained single connected area with a smooth contour is regarded as the effective spraying surface image of each partially covered component.

[0160] In this embodiment, the maximum single-side size of the foreground area is the length of the longest partial contour in the contour line of the foreground area, wherein the partial contour is a partial curved contour line or a straight contour line divided by an inflection point.

[0161] In this embodiment, based on the maximum single side size of the foreground area of ​​each valid binary image of each partially covered component, the reasonable structure element size is determined by taking the product of the maximum single side size of the foreground area of ​​each valid binary image of each partially covered component and a preset multiple as the side length x of the reasonable structure element size;

[0162] The preset multiple is a value set according to the big data analysis result, for example, it can be 1, for example, the size of a reasonable structural element is a rectangular range of x*x;

[0163] In this embodiment, an opening operation is performed on each valid binary image of each partially covered component based on a reasonable structure element size until each valid binary image becomes a single connected region image, wherein:

[0164] The opening operation is to first erode and then dilate the effective binary image;

[0165] The basic principle of dilation is to compare a structural element (a rectangle or circle composed of a group of pixels, where the size of the rectangle or circle is the size of the structural element) with each pixel in the image. If the value of the pixel is 1, the position of the central pixel of the structural element is marked as 1; the erosion process is the opposite;

[0166] The process of "corrosion followed by expansion" is continuously and repeatedly performed on the effective binary images until the latest effective binary image becomes a single connected region image, and then the spraying region image corresponding to each effective binary image is obtained.

[0167] In this embodiment, the appearance frequency ratio of a certain pixel point is the ratio of “the total number of all spraying area images containing the certain pixel point” to “the total number of all spraying area images”.

[0168] In this embodiment, the preset frequency ratio threshold is a value that the appearance frequency ratio of the necessary pixel points of the partially covered component cannot be less than.

[0169] In this embodiment, the necessary spraying area of ​​each partially covered component is expanded until a single-connected area with a smooth contour is obtained. The size of the structural element used in the expansion process is the same as the size of the structural element used in the "opening operation" process in the previous step.

[0170] The beneficial effect of the above technology is: by performing an opening operation on all valid binary images of the partially covered parts to obtain the spray area image corresponding to each valid binary image, the necessary spray area of ​​the partially covered parts is reconstructed, and then an expansion process is performed, thereby avoiding local omissions of the necessary spray area of ​​the partially covered components, and greatly ensuring the accuracy of the effective spray surface image of the partially covered components that is finally determined.

[0171] Embodiment 9:

[0172] Based on Example 1, a method for planning the spraying trajectory of automobile covering parts based on machine vision, S3: determining the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component, referring to Figure 6 ,include:

[0173] S301: determining an effective spraying area and an effective spraying surface contour shape of each partially covered component based on an effective spraying surface contour of each partially covered component;

[0174] S302: Determine a planned spraying trajectory of each partially covered component based on the effective spraying area and the effective spraying surface contour shape and the spraying range radius of each partially covered component.

[0175] In this embodiment, the effective spraying area of ​​the partial covering component is the actual area of ​​the region enclosed by the effective spraying surface contour of the partial covering component.

[0176] In this embodiment, the effective spraying surface contour shape of the partial covering component is a shape formed by the effective spraying surface contour of the partial covering component.

[0177] In this embodiment, the spraying range radius is the radius of the effective spraying circle range formed when the spray gun used for spraying directly sprays the spraying surface without moving.

[0178] In this embodiment, based on the effective spraying area and the effective spraying surface contour shape and the spraying range radius of each partially covered component, the planned spraying trajectory of each partially covered component is determined, including:

[0179] Obtaining the effective spraying area and the effective spraying surface contour shape and the spraying range radius of a large number of different partially covered components, as well as the manually planned spraying paths of the corresponding partially covered components;

[0180] The effective spraying area and the effective spraying surface contour shape and the spraying range radius of a large number of partially covered components are used as model inputs, and the manually planned spraying paths of the corresponding partially covered components are used as model outputs to train the neural network model and obtain a spraying trajectory planning model;

[0181] The effective spraying area, effective spraying surface contour shape and spraying range radius of each partially covered component are input into the spraying trajectory planning model to obtain the planned spraying trajectory of each partially covered component (e.g. Figure 6 ).

[0182] The beneficial effect of the above technology is: based on the determined effective spraying area and effective spraying surface contour shape and spraying range radius of each partially covered component, and combined with the neural network model, the planned spraying trajectory of each partially covered component is reasonably determined.

[0183] Embodiment 10:

[0184] The present invention provides a system for planning the spraying trajectory of automobile panels based on machine vision, which is used to execute the method for planning the spraying trajectory of automobile panels based on machine vision described in any one of embodiments 1 to 9, referring to Figure 7 ,include:

[0185] An image processing module is used to perform denoising and enhancement processing on the complete spray surface images of all partially covered components in the automobile covering parts to obtain effective spray surface images of all partially covered components;

[0186] A contour determination module, used for performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component;

[0187] The trajectory planning module is used to determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

[0188] The beneficial effects of the above technology are: by denoising and enhancing the complete spray surface images of all partial covering components in the automobile covering parts, the effective spray surface images of all partial covering components are obtained, which ensures the accuracy and operability of the spray trajectory planning of the covering parts, and based on the effective spray surface contours in the effective spray surface images of all partial covering components, the spray trajectory of all partial covering components in the automobile covering parts is reasonably planned, thereby ensuring the spraying quality of the automobile covering parts.

[0189] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for planning the spraying trajectory of automobile panels based on machine vision, characterized in that: include: S1: De-noising and enhancing the complete spray surface images of all partially covered components of the automobile covering parts to obtain effective spray surface images of all partially covered components; S2: performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component; S3: Determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

2. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 1, characterized in that: S1: De-noise and enhance the complete spray surface images of all partially covered components in the automobile covering parts to obtain the effective spray surface images of all partially covered components, including: S101: Acquire complete spray surface images of all partial covering components in the automobile covering part; S102: performing multiple hypothetical grayscale processing on the complete spray surface image of each partially covered component based on a weighted average method and screening the processing results to obtain all valid hypothetical spray surface grayscale images of each partially covered component; S103: performing multiple assumption filtering processes on all valid assumption spray surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spray surface filtering images of each partially covered component; S104: performing multiple hypothetical binarization processing, white spot and burr elimination, and processing result screening on all valid hypothetical spray surface filter images of each partially covered component to obtain a valid spray surface image of each partially covered component.

3. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 2, characterized in that: S102: Perform multiple hypothetical grayscale processing on the complete spray surface image of each partially covered component based on the weighted average method and screen the processing results to obtain all valid hypothetical spray surface grayscale images of each partially covered component, including: Based on the constraint that the sum of the weight values ​​corresponding to the three channel values ​​is 1, a preset number of groups of hypothetical weight values ​​are generated; The sum of the products of the three channel values ​​of each pixel point in the complete spray surface image of each partially covered component and the corresponding preset weight value in each set of preset weight values ​​is regarded as the grayscale value of each pixel point in the complete spray surface image of each partially covered component relative to the corresponding set of assumed weight values; Generate a hypothetical grayscale image of each partially covered component relative to each set of hypothetical weight values ​​based on the grayscale value of each pixel in the complete spray surface image of each partially covered component relative to each set of hypothetical weight values; Among all the hypothetical grayscale images of each partially covered component, all valid hypothetical spray surface grayscale images of each partially covered component are screened out.

4. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 3 is characterized in that: Among all the assumed grayscale images of each partially covered component, all valid assumed spray surface grayscale images of each partially covered component are screened out, including: The average of the grayscale values ​​of all pixels in each hypothetical grayscale image of each partially covered component is regarded as the overall brightness of each hypothetical grayscale image; All the hypothetical grayscale images of each partially covered component whose overall brightness is not less than the too dark threshold and not greater than the too bright threshold are regarded as all the valid hypothetical spray surface grayscale images of each partially covered component.

5. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 2, characterized in that: S103: performing multiple assumption filtering processes on all valid assumption spray surface grayscale images of each partially covered component and screening the processing results to obtain all valid assumption spray surface filtering images of each partially covered component, including: Performing linear mean filtering on all valid hypothetical spray surface grayscale images of each partially covered component to obtain a first hypothetical filtered image of each valid hypothetical spray surface grayscale image of each partially covered component; Performing median filtering on all valid hypothetical sprayed surface grayscale images of each partially covered component to obtain a second hypothetical filtered image of each valid hypothetical sprayed surface grayscale image of each partially covered component; Inputting the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image into the image clarity evaluation model respectively, obtaining the clarity evaluation value of the first assumed filtered image and the second assumed filtered image of each valid assumed sprayed surface grayscale image; Inputting the first assumption filtered image and the second assumption filtered image of each valid assumption spray surface grayscale image into the edge information integrity evaluation model respectively, obtaining the edge information integrity evaluation value of the first assumption filtered image and the edge information integrity evaluation value of the second assumption filtered image of each valid assumption spray surface grayscale image; Based on the clarity evaluation value and edge information integrity evaluation value of the first hypothetical filtered image of all valid hypothetical spray surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second hypothetical filtered image, all valid hypothetical spray surface filter images of each partially covered component are obtained.

6. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 5, characterized in that: Based on the clarity evaluation value and edge information integrity evaluation value of the first assumed filtered image of all valid assumed sprayed surface grayscale images of each partially covered component and the clarity evaluation value and edge information integrity evaluation value of the second assumed filtered image, all valid assumed sprayed surface filter images of each partially covered component are obtained, including: The first hypothetical filtered image or the second hypothetical filtered image having the largest weighted sum of the clarity evaluation value and the edge information integrity evaluation value among the first hypothetical filtered image and the second hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component is regarded as the third hypothetical filtered image of each effective hypothetical sprayed surface grayscale image of each partially covered component; All third hypothetical filtered images of each partially covered component, whose weighted sum of clarity evaluation value and edge information integrity evaluation value is not less than a threshold, are regarded as all valid hypothetical spray surface filter images of each partially covered component.

7. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 2, characterized in that: S104: performing multiple hypothetical binarization processing, white spot and burr elimination, and processing result screening on all effective hypothetical spray surface filter images of each partially covered component to obtain an effective spray surface image of each partially covered component, including: Based on the Otsu algorithm, all valid hypothetical spray surface filter images of each partially covered component are binarized to obtain all hypothetical binary images of each partially covered component; Determine the region composed of all pure black pixels in each assumed binary image as the first region of each assumed binary image, and at the same time, determine the region composed of all pure white pixels in each assumed binary image as the second region of each assumed binary image; The first area or the second area with the largest number of pixels in each assumed binary image is regarded as a foreground area, and the first area or the second area with the least number of pixels in each assumed binary image is regarded as a background area; The ratio of the total number of pixels in the background area to the total number of pixels in the foreground area in each hypothetical binary image is regarded as the validity of each hypothetical binary image, and all hypothetical binary images whose validity is not less than the validity threshold among all hypothetical binary images of each partially covered component are regarded as all valid binary images of each partially covered component; All effective binary images of each partially covered component are subjected to white spot and burr elimination and processing result screening to obtain an effective spray surface image of each partially covered component.

8. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 7, characterized in that: All valid binary images of each partially covered component are subjected to white spot and burr elimination and processing result screening to obtain the valid spray surface image of each partially covered component, including: Determine a reasonable structure element size based on the maximum single-side size of the foreground area of ​​each valid binary image of each partially covered component; Based on the reasonable structure element size, each effective binary image of each partially covered component is opened until each effective binary image becomes a single connected region image, and a spraying region image corresponding to each effective binary image is obtained; The overlapping area of ​​the spraying area images corresponding to all valid binary images of each partially covered component is regarded as the necessary area of ​​each partially covered component, and all pixel points except the necessary area in each spraying area image of the corresponding partially covered component are regarded as all accidental pixel points of the corresponding spraying area image; In all spraying area images of each partially covered component, the occasional pixel points whose appearance frequency ratio is not less than the preset frequency ratio threshold are regarded as the inevitable pixel points of the corresponding partially covered component, and the inevitable area and all inevitable pixel points of each partially covered component are summarized to obtain the inevitable spraying area of ​​each partially covered component; The necessary spraying area of ​​each partially covered component is expanded until a single connected area with a smooth contour is obtained, and the obtained single connected area with a smooth contour is regarded as the effective spraying surface image of each partially covered component.

9. The method for planning the spraying trajectory of automobile panels based on machine vision according to claim 1, characterized in that: S3: Determining a planned spraying trajectory of each partially covered component based on the effective spraying surface profile of each partially covered component, including: S301: determining an effective spraying area and an effective spraying surface contour shape of each partially covered component based on an effective spraying surface contour of each partially covered component; S302: Determine a planned spraying trajectory of each partially covered component based on the effective spraying area and the effective spraying surface contour shape and the spraying range radius of each partially covered component.

10. The automobile panel spraying trajectory planning system based on machine vision is characterized by: The method for planning the spraying trajectory of automobile panels based on machine vision according to any one of claims 1 to 9 comprises: An image processing module is used to perform denoising and enhancement processing on the complete spray surface images of all partially covered components in the automobile covering parts to obtain effective spray surface images of all partially covered components; A contour determination module, used for performing contour recognition on the effective spray surface image of each partially covered component to obtain the effective spray surface contour of each partially covered component; The trajectory planning module is used to determine the planned spraying trajectory of each partially covering component based on the effective spraying surface profile of each partially covering component.

Images