Computer-aided aluminum veneer polishing quality assessment method and system

Abstract

The present invention relates to a computer-aided aluminum veneer polishing quality assessment method and system, and belongs to the field of electronic data digital processing technology. The method utilizes a computer for auxiliary design and processing, and the method can be applied to new software and new technical services. The method obtains a first characteristic index corresponding to each characteristic area according to each texture structure characteristic index; obtains a second characteristic index corresponding to each characteristic area according to an R channel intensity difference index, a G channel intensity difference index, and a B channel intensity difference index; and evaluates the polishing quality corresponding to the polishing process when producing aluminum veneers according to the first characteristic index and the second characteristic index. Compared with the method of evaluating the polishing quality of aluminum veneers based on manual visual inspection, the present invention can more reliably evaluate the polishing quality corresponding to the polishing process when producing aluminum veneers, and the work efficiency of evaluating the polishing quality corresponding to the polishing process is also high.

Description

Technical Field

[0001] The invention relates to the technical field of electronic data digital processing, and in particular to a computer-aided aluminum veneer grinding quality assessment method and system. Background Art

[0002] Aluminum veneer is a kind of decoration material and also a new type of material. Aluminum veneer is used in indoor aluminum veneer ceilings, building curtain walls, columns and beams, balconies, partition decorations, interior decoration, etc. in various buildings such as shopping malls, hotels, airports, railway stations, subways, office buildings, commercial and residential buildings, etc. Exterior wall aluminum veneer is also used for thermal insulation of building exterior walls. The sheet metal used to make aluminum veneer generally needs to go through the processes of shearing, punching, folding, welding, nailing and reinforcement, grinding, polishing, spraying and baking to form aluminum veneer. However, during the sheet metal grinding process, scratches, grinding marks or cracks may appear on the surface of the sheet metal due to operation or machine reasons. This will not only affect the appearance of the produced aluminum veneer, but in severe cases, it will reduce the effect of the aluminum veneer and affect its service life. Therefore, the inspection and evaluation of the sheet metal grinding quality of aluminum veneer is a very important link.

[0003] The existing method of evaluating the grinding quality of aluminum veneers is generally based on manual visual inspection. This method is highly subjective, labor-intensive, and has low work efficiency. In addition, missed detection or false detection may occur during the detection and evaluation process. Therefore, this method has low reliability in evaluating the grinding quality of aluminum veneers. Summary of the invention

[0004] The present invention provides a computer-aided aluminum veneer grinding quality assessment method and system, which is used to solve the problem that the existing aluminum veneer grinding quality cannot be reliably assessed. The technical solution adopted is as follows:

[0005] In a first aspect, an embodiment of the present invention provides a computer-aided aluminum veneer polishing quality assessment method and system, comprising the following steps:

[0006] Acquire a sheet metal target image after a grinding process when producing an aluminum single plate; the sheet metal target image is an RGB image;

[0007] According to the pixel value of each pixel point in the sheet metal target image, each feature area corresponding to the sheet metal target image is obtained;

[0008] According to the neighboring pixel points of each pixel point in each feature area in each feature direction, each texture structure feature index corresponding to each feature area is obtained; according to each texture structure feature index, a first feature index corresponding to each feature area is obtained;

[0009] According to the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area, an R channel intensity difference index, a G channel intensity difference index, and a B channel intensity difference index corresponding to each feature area are obtained;

[0010] According to the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index, a second characteristic index corresponding to each characteristic region is obtained;

[0011] According to the first characteristic index and the second characteristic index, the polishing quality corresponding to the polishing process when producing the aluminum veneer is evaluated.

[0012] The present invention also provides a computer-aided aluminum veneer polishing quality assessment system, comprising a memory and a processor, wherein the processor executes a computer program stored in the memory to implement the above-mentioned computer-aided aluminum veneer polishing quality assessment method.

[0013] Beneficial effects: The present invention uses the neighboring pixel points of each pixel point in each feature area in each feature direction as the basis for obtaining each texture structure feature index corresponding to each feature area; uses each texture structure feature index as the basis for obtaining the first feature index corresponding to each feature area; uses the values ​​of the three channels of R, G, and B corresponding to each pixel point in each feature area as the basis for obtaining the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index corresponding to each feature area; uses the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index as the basis for obtaining the second feature index corresponding to each feature area; uses the first feature index and the second feature index as the basis for evaluating the polishing quality corresponding to the polishing process when producing aluminum veneers. The method uses a computer for auxiliary design and processing; and the method can be applied to new software and new technical services. Compared with the method of evaluating the polishing quality of aluminum veneers based on manual visual inspection, the present invention can more reliably evaluate the polishing quality corresponding to the polishing process when producing aluminum veneers, and the work efficiency of evaluating the polishing quality corresponding to the polishing process is also high.

[0014] Preferably, the method for obtaining each feature area corresponding to the sheet metal target image according to the pixel value of each pixel point in the sheet metal target image includes:

[0015] grayscale the sheet metal target image to obtain a sheet metal grayscale image corresponding to the sheet metal target image;

[0016] The DBSCAN clustering algorithm is used to cluster the pixels on the sheet metal grayscale image;

[0017] Obtaining a category with the largest number of pixels, and recording areas corresponding to other categories except the category with the largest number of pixels as characteristic areas corresponding to the sheet metal grayscale image;

[0018] According to the characteristic regions corresponding to the sheet metal grayscale image, the characteristic regions corresponding to the sheet metal target image are obtained.

[0019] Preferably, the method of obtaining each texture structure feature index corresponding to each feature area according to the neighboring pixel points of each pixel point in each feature direction in each feature area includes:

[0020] Obtain each feature direction corresponding to the sheet metal target image and the angle corresponding to each feature direction;

[0021] Obtaining neighboring pixel points of each pixel point in each feature direction in each feature area;

[0022] Calculate the average grayscale value of the neighboring pixels of each pixel in each feature area in each feature direction to obtain the grayscale average value of each pixel in each feature area in each feature direction;

[0023] Constructing a feature descriptor corresponding to each pixel in each feature region in each feature direction according to the grayscale value corresponding to each pixel in each feature region and the grayscale mean value of each pixel in each feature region in each feature direction;

[0024] Acquire each feature descriptor corresponding to each feature region in each feature direction and the number of types of feature descriptors corresponding to each feature region in each feature direction;

[0025] According to the feature descriptors corresponding to the feature regions in the feature directions and the number of types of feature descriptors corresponding to the feature regions in the feature directions, the probability of each type of feature descriptor of each feature region in the feature directions being in the feature descriptors corresponding to the corresponding feature regions in the feature directions is obtained;

[0026] According to the probability, a texture structure feature index corresponding to each feature area is obtained.

[0027] Preferably, the method of obtaining the first characteristic index corresponding to each characteristic area according to each texture structure characteristic index includes:

[0028] Obtaining the center point position corresponding to each feature area;

[0029] Draw a straight line through the center point corresponding to each characteristic region at the same angle as that corresponding to each characteristic direction, and record the straight line as the first straight line corresponding to each characteristic region in each characteristic direction;

[0030] Obtain each pixel point in the feature region corresponding to the first straight line corresponding to each feature region in each feature direction, and record each pixel point in the feature region corresponding to the first straight line corresponding to each feature region in each feature direction as each first center point corresponding to each feature region in each feature direction;

[0031] Draw a straight line perpendicular to the first straight line corresponding to the corresponding feature region in each feature direction through each first center point corresponding to each feature region in each feature direction, to obtain a vertical line corresponding to each first center point corresponding to each feature region in each feature direction;

[0032] According to the pixel points on the vertical line, a pixel point sequence corresponding to each first central point corresponding to each feature area in each feature direction is constructed;

[0033] Constructing a mixed Gaussian model corresponding to each first center point of each feature region in each feature direction according to the grayscale value in the pixel point sequence;

[0034] Calculate the KL divergence between the mixed Gaussian models corresponding to any two first center points of each feature region in each feature direction to obtain each first divergence corresponding to each feature region in each feature direction;

[0035] The first divergences corresponding to the feature regions in the feature directions are summed and then averaged to obtain the first target divergences corresponding to the feature regions in the feature directions;

[0036] According to the first target divergence, the reliability of each texture structure feature index corresponding to each feature area in each feature direction is calculated;

[0037] According to the reliability of each texture structure characteristic index and each texture structure characteristic index, obtaining each target texture structure characteristic index corresponding to each characteristic area;

[0038] According to the target texture structure feature indicators, a first feature indicator corresponding to each feature area is obtained.

[0039] Preferably, the reliability of each texture structure feature index corresponding to each feature area in each feature direction is calculated according to the following formula:

[0040]

[0041] Among them, C i,m C1 is the reliability of the texture structure feature index corresponding to the i-th feature area in the m-th feature direction, i,m is the first target divergence corresponding to the i-th feature region in the m-th feature direction, w is the setting parameter, and w1 is the setting parameter.

[0042] Preferably, according to the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area, a method for obtaining an R channel intensity difference index, a G channel intensity difference index, and a B channel intensity difference index corresponding to each feature area includes:

[0043] Get the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area and the central pixel point corresponding to each feature area;

[0044] According to the mean value of the pixel value of each pixel point in each feature area corresponding to the R channel, the mean value of the pixel value of each pixel point in each feature area corresponding to the G channel, and the mean value of the pixel value of each pixel point in each feature area corresponding to the B channel, the R channel intensity, the G channel intensity, and the B channel intensity corresponding to each feature area are obtained;

[0045] According to the difference between the R channel intensities corresponding to any two feature areas, the difference between the G channel intensities, the difference between the B channel intensities and the central pixel points in any two feature areas, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area are obtained.

[0046] Preferably, the R channel intensity difference index corresponding to each feature area is calculated according to the following formula:

[0047]

[0048] Among them, S i R is the R channel intensity difference index corresponding to the i-th feature region in the sheet metal target image, I is the number of feature regions in the sheet metal target image, and L i,j is the Euclidean distance between the central pixel point corresponding to the i-th feature area in the sheet metal target image and the central pixel point corresponding to the j-th feature area, T is the setting parameter, σ is the setting parameter, H i R is the R channel intensity corresponding to the i-th feature area in the sheet metal target image, H j R is the R channel intensity corresponding to the jth feature area in the sheet metal target image.

[0049] Preferably, the method for obtaining the second characteristic index corresponding to each characteristic region according to the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index includes:

[0050] Calculate the sum of the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index corresponding to each characteristic region to obtain a comprehensive difference index corresponding to each characteristic region;

[0051] According to the comprehensive difference index, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each characteristic region, a second characteristic index corresponding to each characteristic region is obtained.

[0052] Preferably, the method for evaluating the polishing quality corresponding to the polishing process when producing the aluminum single plate according to the first characteristic index and the second characteristic index includes:

[0053] Multiplying the first characteristic index corresponding to each characteristic region by the corresponding second characteristic index to obtain a comprehensive characteristic index corresponding to each characteristic region;

[0054] Obtaining the number of pixels in each feature area;

[0055] Multiplying the comprehensive feature index corresponding to each feature area by the number of pixels in the corresponding feature area to obtain the abnormality degree corresponding to each feature area;

[0056] Sum the abnormality levels corresponding to the characteristic regions to obtain the abnormality level corresponding to the sheet metal target image;

[0057] According to the abnormality degree corresponding to the sheet metal target image, the polishing quality corresponding to the polishing process when producing the aluminum single plate is evaluated. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 The present invention is a flowchart of a computer-aided aluminum veneer grinding quality assessment method. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the embodiments of the present invention.

[0061] 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 invention pertains.

[0062] This embodiment provides a computer-aided aluminum plate polishing quality assessment method, which is described in detail as follows:

[0063] like Figure 1 As shown, the computer-aided aluminum veneer grinding quality assessment method includes the following steps:

[0064] Step S001, obtaining a sheet metal target image after a grinding process when producing an aluminum single plate; the sheet metal target image is an RGB image.

[0065] In this embodiment, the aluminum veneer processing process is generally divided into two steps: the first step is sheet metal processing, and the second step is spraying. The sheet metal processing process mainly processes the sheet metal into the shape and size required for construction through cutting, folding, bending, welding, grinding and other processes; and the grinding process in the sheet metal processing process may cause scratches, grinding marks or cracks on the sheet metal surface due to operation or machine reasons. When such defects are serious, they may affect the use effect of the produced aluminum veneer; considering that when defects appear on the sheet metal surface after grinding, the texture structure of the defective area will change and no longer have the same periodicity as the normal area, this embodiment analyzes the sheet metal image after the grinding process in the processing of the aluminum veneer, realizes the evaluation of the grinding quality in the processing of the aluminum veneer, and provides corresponding reference opinions to relevant staff.

[0066] In this embodiment, the camera is deployed directly above the sheet metal of the aluminum veneer. The camera captures the surface image of the sheet metal after the polishing process when producing the aluminum veneer from a top-down perspective, which can prevent the influence of the camera perspective on image acquisition and improve accuracy; and the shooting range and angle of the camera need to be adjusted according to actual conditions; but the camera shooting range can cover the surface of the sheet metal of the aluminum veneer so as to conduct a comprehensive inspection of it.

[0067] Since the processing environment of aluminum veneer production is relatively complex and there are many noise influences in the environment, the present embodiment performs image preprocessing operation on the sheet metal surface image after the grinding process, that is, the sheet metal surface image after the grinding process is filtered by a Gaussian filtering algorithm to obtain a denoised filtered sheet metal surface image, and the filtering process can eliminate noise points in the sheet metal surface image; and because the surface of the aluminum veneer has the characteristic of light reflection, the present embodiment uses histogram equalization to perform light uniformization processing on the denoised filtered sheet metal surface image to avoid the influence of uneven light on the subsequent grinding quality evaluation; the filtered sheet metal surface image after light uniformization processing is recorded as the sheet metal target image after the grinding process when producing aluminum veneer; the sheet metal target image is an RGB image.

[0068] Step S002: obtaining each feature area corresponding to the sheet metal target image according to the pixel value of each pixel point in the sheet metal target image.

[0069] In this embodiment, when there is a defective area on the sheet metal target image, the texture structure of the defective area will change relative to the normal area; therefore, it is necessary to analyze the pixel value of each pixel point on the sheet metal target image to obtain the feature areas corresponding to the sheet metal target image; the feature areas corresponding to the sheet metal target image are used as the basis for subsequent analysis of the polishing quality corresponding to the sheet metal target image.

[0070] In this embodiment, the sheet metal target image is grayed to obtain a sheet metal grayscale image corresponding to the sheet metal target image; the DBSCAN clustering algorithm is used to cluster each pixel on the sheet metal grayscale image to obtain multiple categories, the number of categories is recorded as U, and the clustering radius needs to be set according to actual conditions; and the features of each pixel in the same category are mostly similar, that is, the grayscale values ​​of each pixel in the same category are similar; since defective areas do not exist in large quantities on the sheet metal grayscale image, the category with the largest number of pixels is obtained, and the category with the largest number of pixels is the normal area, and the areas corresponding to other categories except the category with the largest number of pixels are recorded as the characteristic areas corresponding to the sheet metal grayscale image, and the characteristic areas may be areas where flaws or defects exist, and the characteristic areas need to be further analyzed later; since the sheet metal target image and each position on the sheet metal grayscale image are in a one-to-one correspondence, the characteristic areas corresponding to the sheet metal target image are obtained according to the characteristic areas corresponding to the sheet metal grayscale image.

[0071] Step S003, obtaining each texture structure feature index corresponding to each feature area according to the neighboring pixel points of each pixel point in each feature direction in the feature area; and obtaining the first feature index corresponding to each feature area according to the texture structure feature index.

[0072] In this embodiment, in order to analyze the impact of the grinding process on the sheet metal surface during the processing of aluminum veneer, that is, the quality of grinding during the processing of aluminum veneer, it is necessary to analyze the neighborhood pixel points of each pixel point in each feature direction in each feature area to obtain the texture structure feature index corresponding to each feature area; then, according to the texture structure feature index corresponding to each feature area, the first feature index corresponding to each feature area is obtained; the first feature index corresponding to each feature area is used as the basis for the subsequent evaluation of the grinding quality of the grinding process when producing aluminum veneer.

[0073] (a) According to the neighboring pixel points of each pixel point in each feature area in each feature direction, the specific process of obtaining each texture structure feature index corresponding to each feature area is as follows:

[0074] In this embodiment, since the texture information analyzed for each feature area in different directions will be different, in order to make the extracted texture information more accurate, this embodiment analyzes each feature area from multiple directions; the specific process is: obtaining two neighboring pixel points of each pixel point in each feature area corresponding to the sheet metal grayscale image in each feature direction, that is, taking each pixel point in each feature area corresponding to the sheet metal grayscale image as the center, obtaining the left and right neighboring pixel points of each center point in each feature direction corresponding to the sheet metal grayscale image; when each pixel point in each feature area corresponding to the sheet metal grayscale image is an edge pixel point, only one neighboring pixel point of the edge pixel point in each feature direction can be obtained, and in addition If a neighborhood pixel does not exist, this embodiment performs interpolation processing on the non-existent neighborhood pixel points, thereby obtaining two neighborhood pixel points in each feature direction of each pixel point in each feature area corresponding to the sheet metal grayscale image; and calculates the average of the grayscale values ​​of the two neighborhood pixel points in each feature direction of each pixel point in each feature area corresponding to the sheet metal grayscale image to obtain the grayscale mean value of each pixel point in each feature direction in each feature area corresponding to the sheet metal grayscale image; according to the grayscale value corresponding to each pixel point in each feature area corresponding to the sheet metal grayscale image and the grayscale mean value of each pixel point in each feature direction in each feature area corresponding to the sheet metal grayscale image, constructs a feature descriptor corresponding to each pixel point in each feature direction in each feature area corresponding to the sheet metal grayscale image. Among them, G is the feature descriptor corresponding to a pixel point in a certain feature area corresponding to the sheet metal grayscale image in a certain feature direction, g0 is the grayscale value corresponding to the pixel point in the feature area, is the grayscale mean of the pixel point in the feature area in the feature direction.

[0075] In this embodiment, the number of feature directions set is four, namely, the first feature direction is 0 degrees, the second feature direction is 45 degrees, the third feature direction is 90 degrees, and the fourth feature direction is 135 degrees; therefore, the first feature descriptor corresponding to each pixel point in each feature area corresponding to the sheet metal grayscale image in the first feature direction, the second feature descriptor corresponding to each pixel point in each feature area in the second feature direction, the third feature descriptor corresponding to each pixel point in each feature area in the third feature direction, and the fourth feature descriptor corresponding to each pixel point in each feature area in the fourth feature direction can be obtained.

[0076] As other implementations, other numbers of characteristic directions may be set or other angles may be set for the characteristic directions according to different requirements. For example, the number of characteristic directions may be two, and the angles of the characteristic directions may be 30 degrees and 90 degrees.

[0077] In this embodiment, the number of types of first feature descriptors in each feature area corresponding to the sheet metal grayscale image and the probability of each type of first feature descriptor in each feature area in the corresponding feature area are obtained; according to the probability of each type of first feature descriptor in each feature area in the corresponding feature area, the first texture structure feature index corresponding to each feature area in the first feature direction is obtained; the first texture structure feature index corresponding to each feature area in the first feature direction corresponding to the sheet metal grayscale image is calculated according to the following formula:

[0078]

[0079] Among them, B is the first texture structure feature index corresponding to a certain feature area in the first feature direction corresponding to the sheet metal grayscale image, b1 is the number of types of the first feature descriptor in the feature area corresponding to the sheet metal grayscale image, and B1 b is the probability of the first feature descriptor of the bth category in the feature area corresponding to the sheet metal grayscale image in the corresponding feature area; the larger B is, the more chaotic the texture corresponding to the first feature direction of the feature area is; the logarithm in this embodiment takes 2 as the base, and the unit is bit.

[0080] Then, the probability of each type of second feature descriptor in each feature area in the corresponding feature area is obtained; according to the probability of each type of second feature descriptor in each feature area in the corresponding feature area, the second texture structure feature index corresponding to each feature area in the second feature direction is obtained; the probability of each type of third feature descriptor in each feature area in the corresponding feature area is obtained; according to the probability of each type of third feature descriptor in each feature area in the corresponding feature area, the third texture structure feature index corresponding to each feature area in the third feature direction is obtained; the probability of each type of fourth feature descriptor in each feature area in the corresponding feature area is obtained; according to the probability of each type of fourth feature descriptor in each feature area in the corresponding feature area, the fourth texture structure feature index corresponding to each feature area in the fourth feature direction is obtained; the process of obtaining the second texture structure feature index, the third texture structure feature index and the fourth texture structure feature index in this embodiment is the same as the method of calculating the first texture structure feature index corresponding to each feature area in the first feature direction, so this embodiment will not be described in detail.

[0081] (b) According to each texture structure feature index corresponding to each feature area, the specific process of obtaining the first feature index corresponding to each feature area is as follows:

[0082] Since the above-mentioned characteristic directions are artificially set, the reliability of the texture structure characteristic indicators extracted in different directions is different. Therefore, this embodiment analyzes the pixel points on both sides of the characteristic direction to obtain the reliability of each texture structure characteristic indicator corresponding to each characteristic area. Specifically, the center point position corresponding to each feature area is obtained, and then a straight line with the same angle as the first feature direction is drawn through the center point corresponding to each feature area, and the straight line is recorded as the first straight line corresponding to each feature area in the first feature direction; the pixel points in the feature area corresponding to the first straight line corresponding to each feature area in the first feature direction are obtained, and the pixel points in the feature area corresponding to the first straight line corresponding to each feature area in the first feature direction are recorded as the first center points corresponding to each feature area in the first feature direction; then a straight line perpendicular to the first straight line corresponding to each feature area in the first feature direction is drawn through each first center point corresponding to each feature area in the first feature direction to obtain a vertical line corresponding to each first center point corresponding to each feature area in the first feature direction, and N pixel points on the left side of each first center point and N pixel points on the right side of the first center point are selected on the vertical line, and the selected N pixel points on the left side of each first center point and the N pixel points on the right side of the first center point must be in the feature area corresponding to each first center point; according to the N pixels on the left side of each first center point selected on the vertical line, The pixel point sequence corresponding to each first center point of each feature area in the first feature direction is constructed from the N pixel points on the right side of the prime point and the first center point; the mixed Gaussian model corresponding to each first center point of each feature area in the first feature direction is constructed according to the grayscale value in the pixel point sequence corresponding to each first center point of each feature area in the first feature direction; the KL divergence between the mixed Gaussian models corresponding to any two first center points of each feature area in the first feature direction is calculated to obtain each first divergence corresponding to each feature area in the first feature direction; the calculation process of the KL divergence is a prior art, so this embodiment will not be described in detail; the first divergence corresponding to each feature area in the first feature direction is summed and then averaged to obtain the first target divergence corresponding to each feature area in the first feature direction; the reliability of the first texture structure feature index corresponding to each feature area in the first feature direction is calculated according to the first target divergence corresponding to each feature area in the first feature direction; the reliability of the first texture structure feature index corresponding to each feature area in the first feature direction is calculated according to the following formula:

[0083]

[0084] Among them, C i C1 is the reliability of the first texture structure feature index corresponding to the i-th feature area in the first feature direction corresponding to the sheet metal grayscale image, iis the first target divergence corresponding to the first feature direction of the i-th feature area corresponding to the sheet metal grayscale image, w is the setting parameter, w1 is the setting parameter; C i The larger the value is, the higher the reliability of the first texture structure feature index corresponding to the i-th feature area in the first feature direction is; C1 i The larger the C i The bigger.

[0085] In this embodiment, the value ranges of parameters w and w1 are set to (0, 1), the value of w is set to 0.6, and the value of w1 is set to 0.5; as other implementation methods, the specific values ​​of the setting parameters need to be set according to actual conditions.

[0086] In this embodiment, the reliability of the second texture structure feature index corresponding to each feature area in the second feature direction, the reliability of the third texture structure feature index corresponding to each feature area in the third feature direction, and the reliability of the fourth texture structure feature index corresponding to each feature area in the fourth feature direction can be calculated by analogy; the reliability of the second texture structure feature index corresponding to each feature area in the second feature direction, the reliability of the third texture structure feature index corresponding to the third feature direction, and the reliability of the fourth texture structure feature index corresponding to the fourth feature direction are calculated in the same way as the reliability of the first texture structure feature index corresponding to each feature area in the first feature direction, so this embodiment will not be described in detail.

[0087] In this embodiment, the result of multiplying the reliability of the first texture structure feature indicator corresponding to each feature area in the first feature direction by the corresponding first texture structure feature indicator is recorded as the first target texture structure feature indicator corresponding to each feature area, and so on, the second target texture structure feature indicator corresponding to each feature area, the third target texture structure feature indicator corresponding to each feature area and the fourth target texture structure feature indicator corresponding to each feature area are obtained; the first target texture structure feature indicator, the second target texture structure feature indicator, the third target texture structure feature indicator and the fourth target texture structure feature indicator corresponding to each feature area are summed, and the summation result is recorded as the first feature indicator corresponding to each feature area; the larger the value of the first feature indicator corresponding to each feature area, the more complicated and messy the texture information in the corresponding feature area is, the more scattered the distribution is, and the higher the degree of abnormality in the corresponding feature area is, the more serious the defect is.

[0088] Step S004, according to the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area, obtain the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index corresponding to each feature area.

[0089] In this embodiment, since the first characteristic index corresponding to each of the above-mentioned feature areas reflects texture information, and when the texture distribution in the defect area is relatively uniform, the reliability of relying solely on the first characteristic index to reflect the quality of the sheet metal area corresponding to the sheet metal target image is low, that is, the reliability of evaluating the polishing quality in the processing of the aluminum veneer is low, and therefore it is necessary to analyze the degree of color difference between each feature area to further reflect the degree of abnormality of each feature area; therefore, this embodiment will analyze the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area to obtain the R channel intensity difference index, G channel intensity difference index, and B channel intensity difference index corresponding to each feature area; the obtained R channel intensity difference index, G channel intensity difference index, and B channel intensity difference index corresponding to each feature area will be used as the basis for subsequent analysis to obtain the second characteristic index corresponding to each feature area.

[0090] In this embodiment, the sheet metal target image is channel separated to obtain image data of the three channels of R, G, and B corresponding to each feature area, that is, the values ​​of the three channels of R, G, and B corresponding to each pixel in each feature area are obtained; the mean of the pixel values ​​of each pixel in each feature area corresponding to the R channel is calculated, the mean of the pixel values ​​of each pixel in each feature area corresponding to the G channel is calculated, and the mean of the pixel values ​​of each pixel in each feature area corresponding to the B channel is calculated; the mean of the pixel values ​​of each pixel in each feature area corresponding to the R channel, the mean of the pixel values ​​of each pixel in each feature area corresponding to the G channel, and the mean of the pixel values ​​of each pixel in each feature area corresponding to the B channel are respectively recorded as the R channel intensity, G channel intensity, and B channel intensity corresponding to each feature area.

[0091] According to the difference between the R channel intensities, the G channel intensities, the B channel intensities and the central pixel points in any two feature areas, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area are obtained; the R channel intensity difference index corresponding to each feature area is calculated according to the following formula:

[0092]

[0093] Among them, S i R is the R channel intensity difference index corresponding to the i-th feature region in the sheet metal target image, I is the number of feature regions in the sheet metal target image, and L i,j is the Euclidean distance between the central pixel point corresponding to the i-th feature area in the sheet metal target image and the central pixel point corresponding to the j-th feature area, T is the setting parameter, σ is the setting parameter, H i R is the R channel intensity corresponding to the i-th feature area in the sheet metal target image, Hj R is the R channel intensity corresponding to the jth feature area in the sheet metal target image; when S i R The larger the value is, the greater the difference between the R channel values ​​corresponding to the pixels in the i-th feature area and the R channel values ​​corresponding to the pixels in other feature areas. That is, the greater the color difference between the i-th feature area and the other feature areas, the greater the color difference.

[0094] In this embodiment, the value of the setting parameter T is set to 2, and the value of the setting parameter σ is set to 5. As other implementations, other values ​​may be set for the setting parameters according to different actual conditions.

[0095] In this embodiment, the R channel intensity difference index corresponding to each feature area can be obtained through the above process, and similarly, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area can be obtained; the calculation method of the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area is the same as the calculation method of the R channel intensity difference index corresponding to each feature area, so they will not be described in detail.

[0096] Step S005: obtaining a second characteristic index corresponding to each characteristic region according to the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index.

[0097] In this embodiment, since the first characteristic index corresponding to the characteristic area reflects texture information, and the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index reflect color information, the larger the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each characteristic area are, the higher the degree of color difference between the corresponding characteristic area and other characteristic areas, that is, the higher the degree of abnormality of the corresponding characteristic area; therefore, this embodiment obtains the second characteristic index corresponding to each characteristic area according to the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index; the second characteristic index corresponding to each characteristic area is used as the basis for subsequent evaluation of the polishing quality corresponding to the polishing process when producing aluminum veneer.

[0098] The sum of the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area is calculated to obtain the comprehensive difference index corresponding to each feature area; the second feature index corresponding to each feature area is obtained according to the comprehensive difference index, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area; the second feature index corresponding to each feature area is calculated according to the following formula:

[0099]

[0100] in, is the second feature index corresponding to the i-th feature area in the sheet metal target image, S i R is the R channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i G is the G channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i B is the B channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i is the comprehensive difference index corresponding to the i-th feature area in the sheet metal target image; The larger the value is, the higher the color difference between the i-th feature area and other feature areas in the sheet metal target image is, that is, the higher the abnormality of the i-th feature area is.

[0101] Step S006: Evaluate the polishing quality corresponding to the polishing process when producing the aluminum veneer according to the first characteristic index and the second characteristic index.

[0102] In this embodiment, the first characteristic indicator and the second characteristic indicator corresponding to each characteristic area can reflect the abnormality degree of the corresponding characteristic area, and the abnormality degree of the characteristic area can reflect the polishing quality of the polishing process when producing aluminum veneers; therefore, this embodiment achieves the purpose of evaluating the polishing quality of the polishing process when producing aluminum veneers by analyzing the first characteristic indicator and the second characteristic indicator corresponding to each characteristic area.

[0103] In this embodiment, the first characteristic index and the second characteristic index corresponding to each characteristic area are normalized; and the first characteristic index corresponding to each characteristic area after the normalization is multiplied by the corresponding second characteristic index to obtain the comprehensive characteristic index corresponding to each characteristic area; the number of pixels in each characteristic area is obtained; the comprehensive characteristic index corresponding to each characteristic area is multiplied by the number of pixels in the corresponding characteristic area to obtain the abnormality degree corresponding to each characteristic area; the abnormality degrees corresponding to each characteristic area are summed to obtain the abnormality degree corresponding to the sheet metal target image; and the abnormality degree corresponding to the sheet metal target image is normalized to determine whether the abnormality degree corresponding to the sheet metal target image after the normalization is greater than a preset threshold value. If so, it is determined that the grinding quality of the grinding process during the production of aluminum veneers is unqualified, which will affect the appearance and service life of the produced aluminum veneers; the preset threshold value needs to be determined according to actual conditions.

[0104] The computer-aided aluminum veneer polishing quality assessment system of this embodiment includes a memory and a processor, and the processor executes the computer program stored in the memory to implement the above-mentioned computer-aided aluminum veneer polishing quality assessment method.

[0105] Beneficial effects: In this embodiment, the neighboring pixel points of each pixel point in each feature area in each feature direction are used as the basis for obtaining each texture structure feature index corresponding to each feature area; each texture structure feature index is used as the basis for obtaining the first feature index corresponding to each feature area; the values ​​of the three channels of R, G, and B corresponding to each pixel point in each feature area are used as the basis for obtaining the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index corresponding to each feature area; the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index are used as the basis for obtaining the second feature index corresponding to each feature area; the first feature index and the second feature index are used as the basis for evaluating the polishing quality of the polishing process when producing aluminum veneers. This method uses computers for auxiliary design and processing; this method can be applied to new software and new technical services. Compared with the method of evaluating the polishing quality of aluminum veneers based on manual visual inspection, this embodiment can more reliably evaluate the polishing quality corresponding to the polishing process when producing aluminum veneers, and the work efficiency of evaluating the polishing quality corresponding to the polishing process is also high.

[0106] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A computer-aided aluminum veneer polishing quality assessment method, characterized in that: The method comprises the following steps: Acquire a sheet metal target image after a grinding process when producing an aluminum single plate; the sheet metal target image is an RGB image; According to the pixel value of each pixel point in the sheet metal target image, each feature area corresponding to the sheet metal target image is obtained; According to the neighboring pixel points of each pixel point in each feature area in each feature direction, each texture structure feature index corresponding to each feature area is obtained; according to each texture structure feature index, a first feature index corresponding to each feature area is obtained; According to the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area, an R channel intensity difference index, a G channel intensity difference index, and a B channel intensity difference index corresponding to each feature area are obtained; Obtaining a second characteristic index corresponding to each characteristic region according to the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index; According to the first characteristic index and the second characteristic index, the polishing quality corresponding to the polishing process when producing the aluminum single plate is evaluated; The method for obtaining each feature area corresponding to the sheet metal target image according to the pixel value of each pixel point in the sheet metal target image includes: grayscale the sheet metal target image to obtain a sheet metal grayscale image corresponding to the sheet metal target image; The DBSCAN clustering algorithm is used to cluster the pixels on the sheet metal grayscale image; Obtaining a category with the largest number of pixels, and recording areas corresponding to other categories except the category with the largest number of pixels as characteristic areas corresponding to the sheet metal grayscale image; Obtaining, according to each characteristic region corresponding to the sheet metal grayscale image, each characteristic region corresponding to the sheet metal target image; The method of obtaining each texture structure feature index corresponding to each feature area according to the neighboring pixel points of each pixel point in each feature direction in each feature area includes: Obtain each feature direction corresponding to the sheet metal target image and the angle corresponding to each feature direction; Obtaining neighboring pixel points of each pixel point in each feature direction in each feature area; Calculate the average grayscale value of the neighboring pixels of each pixel in each feature area in each feature direction to obtain the grayscale average value of each pixel in each feature area in each feature direction; Constructing a feature descriptor corresponding to each pixel in each feature region in each feature direction according to the grayscale value corresponding to each pixel in each feature region and the grayscale mean value of each pixel in each feature region in each feature direction; Acquire each feature descriptor corresponding to each feature region in each feature direction and the number of types of feature descriptors corresponding to each feature region in each feature direction; According to the feature descriptors corresponding to the feature regions in the feature directions and the number of types of feature descriptors corresponding to the feature regions in the feature directions, the probability of each type of feature descriptor of each feature region in the feature directions being in the feature descriptors corresponding to the corresponding feature regions in the feature directions is obtained; According to the probability, a texture structure characteristic index corresponding to each characteristic area is obtained; The expression of the texture structure characteristic index corresponding to each characteristic area is: Among them, B i,m Represents the texture structure feature index of the i-th feature area in the m-th feature direction; B1 i,m,b b1 represents the probability of the bth feature descriptor in the mth feature direction of the i-th feature region; i,m It represents the number of types of feature descriptors in the i-th feature region in the m-th feature direction; log() represents the logarithmic function with base 2; The method of obtaining the first characteristic index corresponding to each characteristic area according to each texture structure characteristic index includes: Obtaining the center point position corresponding to each feature area; Draw a straight line through the center point corresponding to each characteristic region at the same angle as that corresponding to each characteristic direction, and record the straight line as the first straight line corresponding to each characteristic region in each characteristic direction; Obtain each pixel point in the feature region corresponding to the first straight line corresponding to each feature region in each feature direction, and record each pixel point in the feature region corresponding to the first straight line corresponding to each feature region in each feature direction as each first center point corresponding to each feature region in each feature direction; Draw a straight line perpendicular to the first straight line corresponding to the corresponding feature region in each feature direction through each first center point corresponding to each feature region in each feature direction, to obtain a vertical line corresponding to each first center point corresponding to each feature region in each feature direction; According to the pixel points on the vertical line, a pixel point sequence corresponding to each first central point corresponding to each feature area in each feature direction is constructed; Constructing a mixed Gaussian model corresponding to each first center point of each feature area in each feature direction according to the grayscale value in the pixel point sequence; Calculate the KL divergence between the mixed Gaussian models corresponding to any two first center points of each feature region in each feature direction to obtain each first divergence corresponding to each feature region in each feature direction; The first divergences corresponding to the feature regions in the feature directions are summed and then averaged to obtain the first target divergences corresponding to the feature regions in the feature directions; According to the first target divergence, the reliability of each texture structure feature index corresponding to each feature area in each feature direction is calculated; According to the reliability of each texture structure characteristic index and each texture structure characteristic index, each target texture structure characteristic index corresponding to each characteristic area is obtained; According to each target texture structure characteristic index, obtaining a first characteristic index corresponding to each characteristic area; The expression of the first characteristic index corresponding to each characteristic area is: Among them, K i represents the first characteristic index of the i-th characteristic region; C i,m Indicates the reliability of the texture structure feature index corresponding to the i-th feature area in the m-th feature direction; B i,m Represents the texture structure feature index of the i-th feature area in the m-th feature direction; C1 i,m is the first target divergence corresponding to the ith feature region in the mth feature direction, w is the setting parameter, w1 is the setting parameter; M represents the number of feature directions; C i,m ×B i,m Represents the target texture structure feature index of the i-th feature area in the m-th feature direction; The method of obtaining the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area according to the values ​​of the three channels R, G and B corresponding to each pixel point in each feature area includes: Get the values ​​of the three channels R, G, and B corresponding to each pixel point in each feature area and the central pixel point corresponding to each feature area; According to the mean value of the pixel value of each pixel point in each feature area corresponding to the R channel, the mean value of the pixel value of each pixel point in each feature area corresponding to the G channel, and the mean value of the pixel value of each pixel point in each feature area corresponding to the B channel, the R channel intensity, the G channel intensity, and the B channel intensity corresponding to each feature area are obtained; According to the difference between the R channel intensities, the difference between the G channel intensities, the difference between the B channel intensities and the central pixel points in any two feature areas, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each feature area are obtained; The R channel intensity difference index corresponding to each feature area is calculated according to the following formula: Among them, S i R is the R channel intensity difference index corresponding to the i-th feature region in the sheet metal target image, I is the number of feature regions in the sheet metal target image, and L i,j is the Euclidean distance between the central pixel point corresponding to the i-th feature area in the sheet metal target image and the central pixel point corresponding to the j-th feature area, T is the setting parameter, σ is the setting parameter, H i R is the R channel intensity corresponding to the i-th feature area in the sheet metal target image, H j R is the R channel intensity corresponding to the jth feature area in the sheet metal target image; The method for obtaining the second characteristic index corresponding to each characteristic region according to the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index includes: Calculate the sum of the R channel intensity difference index, the G channel intensity difference index, and the B channel intensity difference index corresponding to each characteristic region to obtain a comprehensive difference index corresponding to each characteristic region; According to the comprehensive difference index, the R channel intensity difference index, the G channel intensity difference index and the B channel intensity difference index corresponding to each characteristic region, a second characteristic index corresponding to each characteristic region is obtained; The expression of the second characteristic index corresponding to each characteristic area is: in, is the second feature index corresponding to the i-th feature area in the sheet metal target image, S i R is the R channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i G is the G channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i B is the B channel intensity difference index corresponding to the i-th feature area in the sheet metal target image, S i is the comprehensive difference index corresponding to the i-th feature area in the sheet metal target image; The method for evaluating the polishing quality corresponding to the polishing process when producing the aluminum single plate according to the first characteristic index and the second characteristic index includes: Multiplying the first characteristic index corresponding to each characteristic region by the corresponding second characteristic index to obtain a comprehensive characteristic index corresponding to each characteristic region; Obtaining the number of pixels in each feature area; Multiplying the comprehensive feature index corresponding to each feature area by the number of pixels in the corresponding feature area to obtain the abnormality degree corresponding to each feature area; Sum the abnormality levels corresponding to the characteristic regions to obtain the abnormality level corresponding to the sheet metal target image; According to the abnormality degree corresponding to the sheet metal target image, the polishing quality corresponding to the polishing process when producing the aluminum single plate is evaluated.

2. A computer-aided aluminum veneer polishing quality assessment system, comprising a memory and a processor, characterized in that: The processor executes the computer program stored in the memory to implement the computer-aided aluminum veneer polishing quality assessment method as described in claim 1.

Images