Washing machine liner plate punching defect detection method

Abstract

The invention relates to the technical field of image processing, in particular to a washing machine liner plate punching defect detection method, which comprises the following steps: acquiring a liner plate punching gray level image to obtain an approximately circular region, screening out a region with higher circularity in the liner plate punching image, analyzing characteristics in the approximately circular region, and detecting the defects of the liner plate punching defect. The method comprises the following steps: acquiring an approximate circle region which accords with the characteristics in a punched hole in the approximate circle region, analyzing the characteristics of high-reflection interference around the approximate circle region, constructing a light ring characteristic coefficient and a light spot characteristic coefficient, and acquiring the approximate circle region which accords with the high-reflection characteristics of the edge of the punched hole, so as to acquire a region with the punching characteristics; according to the method, the area with the punching characteristic is selected for marking, and then the liner plate punching gray level image is segmented by using the watershed algorithm, so that the problem of excessive segmentation of the watershed algorithm is avoided, a more accurate punching area is obtained, and the detection precision of the washing machine liner plate punching defect is improved.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a method for detecting punching defects in an inner tank plate of a washing machine. Background Art

[0002] As an indispensable household appliance in modern families, the stability of the performance and the length of its service life are directly related to the user experience. As one of the core structural parts of the washing machine, the inner plate of the washing machine not only bears the various physical effects of the clothes during the washing process, but also is responsible for the drainage function of the washing machine. Therefore, the punching quality of the inner plate has a vital impact on the drainage performance, structural strength and overall service life of the washing machine.

[0003] The inner liner is usually made of stainless steel because it has good corrosion resistance and sufficient mechanical strength to meet the needs of long-term use of the washing machine. During the production process, the inner liner is stamped to form regularly arranged drainage holes or functional holes. The design of these holes not only affects the drainage efficiency of the washing machine, but also affects the water flow distribution and washing effect inside the washing machine.

[0004] However, during the punching process, due to the influence of various factors, such as mold wear, uneven distribution of material stress, fluctuation of equipment accuracy, etc., the punching quality of the liner plate is often difficult to be fully guaranteed. These factors can easily lead to various defects in the liner plate, such as hole position offset, hole diameter tolerance (i.e., hole diameter greater than or less than the design value), edge burrs, micro cracks, and deformation.

[0005] These punching defects have a significant impact on the performance and service life of the washing machine. Hole position deviation and hole diameter deviation may damage the sealing of the inner liner board, causing water leakage during operation; edge burrs and microcracks may weaken the mechanical properties of the inner liner board, making it a weak link in the structure of the washing machine, prone to fracture or breakage during long-term use; and deformation may change the shape and size of the inner liner board, affecting its matching accuracy with other parts of the washing machine, and thus causing problems such as abnormal noise or structural failure.

[0006] Therefore, punching defect detection has become a key link in the quality control of washing machine production. Traditional defect detection methods, such as visual inspection or the use of ordinary optical inspection equipment, are often easily disturbed by the high reflective properties of the stainless steel surface, resulting in reduced accuracy and efficiency of defect detection. Especially when detecting subtle defects such as edge burrs and microcracks, the highly reflective surface may make these defects difficult to be accurately identified, thus affecting the overall quality control of the washing machine.

[0007] In summary, the existing technology has the problem of insufficient accuracy in detecting punching defects in the inner tank plate of a washing machine, and there is an urgent need for a new technology or method that can overcome the interference of high reflective light on the stainless steel surface and improve the accuracy and efficiency of defect detection. Summary of the invention

[0008] The purpose of the present invention is to provide a method for detecting punching defects in the inner tank plate of a washing machine, so as to solve the problem that the existing defect detection method is easily interfered by the high reflective light on the stainless steel surface, thus affecting the accuracy of defect detection; to this end, the present invention provides the following technical solutions to solve the above problems.

[0009] The present invention provides a method for detecting punching defects of a washing machine inner tank plate, comprising: Acquire the closed edge line of the grayscale image of the punching hole of the inner liner plate to obtain an approximate circular area in the area surrounded by the closed edge line; Dividing the approximate circular area into a suspected light spot area and a light ring area; Calculate the spot characteristic coefficient of each suspected spot area, and take the suspected spot area whose spot characteristic coefficient is greater than or equal to the spot characteristic threshold as the spot area; the spot characteristic coefficient is positively correlated with the contrast and the spot consistency coefficient, and the spot consistency coefficient represents the consistency of the spot appearance position in each suspected spot area; Taking any area of ​​the halo area and the spot area as the target area; calculating the hole characteristic coefficient of the target area, wherein the hole characteristic coefficient is inversely correlated with the variance and mean of the grayscale values ​​of all pixels in the target area; The target area where the hole characteristic coefficient is greater than the hole characteristic threshold is taken as the punching characteristic area; The punching grayscale image of the liner board is segmented with each punching feature area as the center, and defect detection is performed on the segmented grayscale image to obtain the punching defects.

[0010] The above scheme obtains the grayscale image of the punching of the inner liner plate, analyzes the characteristics of the high-reflective interference of the punching on the grayscale image of the punching of the inner liner plate, first obtains the approximate circular area, screens out the area with higher circularity in the image of the punching of the inner liner plate, analyzes the characteristics in the approximate circular area, obtains the approximate circular area that meets the characteristics in the punching within the approximate circular area, and analyzes the characteristics of the high-reflective interference around the approximate circular area, constructs the halo characteristic coefficient and the spot characteristic coefficient, obtains the approximate circular area that is more consistent with the high-reflective characteristics of the punching edge, thereby obtaining the area with the punching feature, selects the area with the punching feature for marking, and then uses the watershed algorithm to segment the grayscale image of the punching of the inner liner plate to avoid the problem of over-segmentation of the watershed algorithm, obtains a more accurate punching area, and improves the detection accuracy of the punching defects of the inner liner plate of the washing machine.

[0011] Optionally, the method of dividing the approximate circular area into the suspected light spot area and the halo area includes: taking the approximate circular area whose halo characteristic coefficient is less than the halo characteristic threshold as the suspected light spot area, and taking the approximate circular area whose halo characteristic coefficient is greater than or equal to the halo characteristic threshold as the halo area; The halo characteristic coefficient hfc is: ;Among them, amp min is the mean value of the gradient amplitude of all pixels on the minimum circumscribed circle of each approximate circular area, amp cir is the mean value of the gradient amplitude of all pixels on the outer circle adjacent to each approximate circle area and the minimum circumscribed circle, var gra is the variance of the grayscale values ​​of all pixels on all outer circles of each approximate circular area, k min is the absolute value of the slope of the minimum fitting straight line of each approximate circular area, k cir is the absolute value of the slope of the minimum adjacent fitting straight line of each approximate circular area; is the preset parameter coefficient, and sigmoid() is the normalization function.

[0012] The above scheme calculates the halo feature coefficient of each approximate circular area through the average situation and dispersion of the grayscale values ​​of the pixel points of the minimum circumscribed circle and the outer circle of the approximate circular area, as well as the slope of the minimum fitting straight line and the minimum adjacent fitting straight line, thereby improving the accuracy of halo feature recognition.

[0013] Optionally, the process of obtaining the minimum fitting straight line includes: For each approximate circular area in the grayscale image of the liner plate, the HOG algorithm is used to obtain the gradient direction histogram of all pixel points on the minimum circumscribed circle of the approximate circular area; Perform straight line fitting on the values ​​of all intervals in the gradient direction histogram to obtain the minimum fitting straight line.

[0014] Optionally, the hole characteristic coefficient is specifically calculated as: ; in, is the hole characteristic coefficient of the target area; var cir is the variance of the grayscale values ​​of all pixels in the target area, ave is the mean of the grayscale values ​​of all pixels in the target area, and ave def is the preset grayscale mean; It is the preset parameter factor.

[0015] The above scheme obtains the edge lines in the grayscale image of the inner liner plate punching, calculates the circularity of the area enclosed by the closed edge lines, and takes the area enclosed by the closed edge lines whose circularity is greater than the preset circularity threshold as the approximate circular area to obtain the in-hole feature coefficient of the approximate circular area, effectively improving the detection accuracy and efficiency, and optimizing the quality control of the inner liner plate punching, thereby improving product reliability and safety.

[0016] Optionally, the approximate circular area is an area enclosed by closed edge lines whose circularity is greater than a preset circularity threshold.

[0017] Optionally, the light spot characteristic coefficient is obtained in the following manner: Among them, ffc is the spot characteristic coefficient of each suspected spot area; ind fac is the spot consistency coefficient of the cluster where the minimum circumscribed circle of each suspected spot area is located, ldc is the light-dark contrast of the minimum circumscribed circle of each suspected spot area; sigmoid() is the normalization function.

[0018] The above scheme obtains the spot characteristic coefficient of the minimum circumscribed circle of each approximate circle area through the spot consistency coefficient and the light-dark contrast of the minimum circumscribed circle of each suspected spot area, thereby improving the accuracy of spot recognition.

[0019] Optionally, the light spot consistency coefficient is specifically: ; Among them, ind fac is the spot consistency coefficient of the cluster; average() is the average function, It is a set of contour coefficients of all feature points in the clustering cluster, and the clustering cluster is obtained by clustering all feature points in the neighborhood of the minimum circumscribed circle of the suspected light spot area with the feature vector of the minimum circumscribed circle of the suspected light spot area as the feature point, and the feature vector is obtained by processing the pixel points in the minimum circumscribed circle of the suspected light spot area using the SIFT algorithm.

[0020] Optionally, the defect acquisition process of the inner liner plate punching includes: The grayscale image of the punching hole of the liner board is segmented using the mark-based watershed algorithm. The area with each punching feature area as the center and a preset radius as the marking area, where the value of the preset radius must be smaller than the radius of the punching hole; The grayscale image of the punching of the liner board in the marked area is segmented to obtain a segmented image. The segmented grayscale image of the punching of the liner board is subjected to defect detection using a template matching algorithm based on an image pyramid to obtain the punching defects of the liner board.

[0021] The above scheme performs image segmentation on the grayscale image of the punching of the liner board in the marked area to obtain the segmented image, and uses a template matching algorithm based on an image pyramid to perform defect detection on the segmented grayscale image of the punching of the liner board to obtain the punching defects of the liner board, thereby improving the accuracy and efficiency of defect detection.

[0022] Optionally, the method for marking the grayscale image of the punching of the liner plate includes: marking the position of the punching feature area on the image using a marking algorithm.

[0023] Optionally, the grayscale image of the punching of the inner liner plate is obtained by capturing the punching image of the inner liner plate using a high-pixel industrial camera, denoising the punching image of the inner liner plate using image denoising, and then grayscale processing is performed on the denoised surface image of the punching of the inner liner plate, wherein the image denoising is performed using a mean filtering denoising algorithm.

[0024] The beneficial effects of the present invention are: The scheme of the present invention can analyze the features of high-reflective interference of punching holes on the grayscale image of the inner liner plate punching by acquiring a grayscale image of the inner liner plate punching. First, an approximate circular area is acquired, and an area with a higher circularity in the image of the inner liner plate punching is screened out. The features in the approximate circular area are analyzed, and an approximate circular area that meets the features in the punching within the approximate circular area is acquired. The features of high-reflective interference around the approximate circular area are analyzed, and halo feature coefficients and spot feature coefficients are constructed to acquire an approximate circular area that is more consistent with the high-reflective features of the punching edge, thereby acquiring an area with punching features, selecting the area with punching features for marking, and then using a watershed algorithm to segment the grayscale image of the inner liner plate punching, thereby avoiding the problem of over-segmentation of the watershed algorithm, acquiring a more accurate punching area, and improving the detection accuracy of punching defects in the inner liner plate of the washing machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flowchart schematically shows a method for detecting punching defects in a washing machine liner panel in this embodiment; Figure 2 A schematic diagram of an outer circle in a method for detecting punching defects of an inner liner plate of a washing machine in this embodiment is schematically shown; Figure 3 The gradient direction histogram in a method for detecting punching defects in an inner liner of a washing machine in this embodiment is schematically shown. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0027] The present invention is directed to a solution for detecting punching defects in an inner tank plate of a washing machine. The solution for detecting punching defects in an inner tank plate needs to utilize an in-hole characteristic coefficient, a halo characteristic coefficient, and a spot characteristic coefficient to obtain a punching feature area, mark an inner tank plate punching grayscale image according to the punching feature area, and segment the inner tank plate punching grayscale image to complete the defect detection of punching holes in the inner tank plate of the washing machine.

[0028] Therefore, based on the above content, the present invention provides a method for detecting punching defects on the inner tank plate of a washing machine to accurately identify the detection of punching defects on the inner tank plate of a washing machine in a highly reflective environment.

[0029] Specifically, Figure 1 As shown, a method for detecting punching defects of a washing machine liner plate in this embodiment includes the following steps: Step S1: collecting the closed edge lines of the grayscale image of the punching holes of the inner liner plate of the washing machine to obtain an approximate circular area in the area surrounded by the closed edge lines.

[0030] Specifically, the present invention selects a steel inner liner plate for defect detection of a washing machine. During the production and processing of the steel inner liner plate of the washing machine, the steel plate is usually used as a flat plate, and regularly distributed drainage holes are punched out on the steel plate. The punched flat plate is then curled into a barrel shape, and the seam is closed by welding. In order to obtain a more accurate image, the present invention uses a high-pixel industrial camera placed on the front to collect the punching surface image of the inner liner plate after regularly distributed drainage holes are punched out on the inner liner plate and before curling it into a cylindrical shape for welding. The image denoising is used to denoise the punching image of the inner liner plate. The denoising algorithm selected in the embodiment of the present invention is a mean filtering denoising algorithm. The implementer can select other image denoising algorithms according to actual conditions, and grayscale the denoised punching surface image of the inner liner plate to obtain a grayscale image of the punching of the inner liner plate.

[0031] In this embodiment, the inner tank plate of the washing machine selected by the present invention is a steel inner tank plate, and the steel inner tank plate is processed by stamping. During the process of punching holes during the stamping process, the inner tank plate usually has a defect of default hole shape.

[0032] Under normal circumstances, the punched holes in the inner liner plate should be circular. When the inner liner plate has defects such as default hole pattern, including burrs, inconsistent hole diameters, hole position offset and shape deformation, the default hole pattern may affect the performance of the washing machine. Abnormal shape and distribution of the punched holes will hinder the rapid discharge of water. Incomplete or uneven distribution of the punched holes will disrupt the normal circulation path of the water flow during washing and reduce the tumbling and friction efficiency of the clothes.

[0033] When detecting the punching defects of stamping parts based on traditional image processing technology, the punching part of the stamping parts is usually obtained based on the image segmentation algorithm, and then the punching defects are detected; however, when the watershed algorithm is used to segment the punching defects of the liner plate, due to the high reflective interference image, reflective areas will appear, resulting in over-segmentation in the image segmentation process, and the punching area cannot be effectively segmented finely. Finally, the segmented punching area is incomplete, which affects the detection accuracy of the punching defects. The present invention analyzes the high reflective features on the punching holes of the washing machine liner plate, and adaptively marks the area with the punching features before using the watershed algorithm to segment the image, so as to obtain a better segmentation effect.

[0034] Based on the above analysis, for the grayscale image of the inner liner plate punching, an edge detection algorithm is used to obtain the edge line in the grayscale image of the inner liner plate punching, and the pixel points on the closed edge line are used as closed edge pixel points.

[0035] For the closed edge line, the circularity of the area enclosed by the closed edge line is calculated. The closer the circularity value is to the value 1, the closer the area enclosed by the closed edge line is to a circle. A preset circularity threshold is set. In one embodiment of the present invention, the preset circularity threshold is 0.9. The implementer may select other values ​​according to actual conditions.

[0036] Specifically, the punching holes on the inner tank plate of the washing machine are in a standard circular shape under normal circumstances. However, since the inner tank plate of the washing machine is relatively smooth, reflective images will be produced, thus affecting the division of the punching area. Therefore, for high reflective interference, specifically: In the inner tank plate punching image, the area with higher circularity may be the punching area on the inner tank plate of the washing machine, or it may be the reflective image on the inner tank plate. Based on the above analysis, the area enclosed by the closed edge lines with a circularity greater than the preset circularity threshold is taken as an approximate circular area, and the area with a circularity greater than the preset circularity threshold is subjected to feature analysis.

[0037] In this embodiment, by collecting the grayscale image of the punching of the inner liner plate of the washing machine, the state of the punching of the inner liner plate can be intuitively displayed, so as to improve the accuracy of the detection of the punching defects of the inner liner plate.

[0038] Step S2: dividing the approximate circular area into a suspected light spot area and a light ring area.

[0039] Specifically, when an image is captured of a punched hole in an inner tank plate of a washing machine, the punched hole and the inner tank plate are not directly cut, but there is a certain curvature, so an aperture or a light spot will appear near the punched hole.

[0040] Therefore, it is necessary to analyze the aperture and spot distribution characteristics of the pixel points punched in the washing machine liner plate. The specific method is as follows: like Figure 2 As shown, for the edge pixel points of each approximate circular area in the grayscale image of the inner liner plate punching, the minimum circumscribed circle that includes the approximate circular area is obtained, and a preset number of adjacent and continuous outer circles of the minimum circumscribed circle are obtained. In one embodiment of the present invention, the preset number is 3, and the implementer can select other values ​​according to actual conditions.

[0041] Aperture features will appear near the punching hole. Since the color inside the punching hole is darker and the edge of the punching hole is not a sharp edge but a gradually transitioning arc edge with a slight bulge, the brightness of the aperture in the convex part of the punching edge area is quite different from that in the punching hole, that is, the grayscale value of the pixel point of the aperture part in the convex part of the punching edge area is extremely large, and the grayscale value of the pixel point in the punching hole is extremely small.

[0042] For each approximate circular area in the grayscale image of the liner plate, the HOG algorithm is used to obtain the gradient direction histogram of all pixels on the minimum circumscribed circle of the approximate circular area, as well as the gradient direction histogram of all pixels on the outer circle adjacent to the minimum circumscribed circle area.

[0043] like Figure 3 As shown, the gradient direction histogram includes the gradient amplitude and gradient direction of the grayscale value of the pixel point; the values ​​of all bins (bin, the interval in the gradient direction histogram) in the gradient direction histogram are subjected to straight line fitting processing, and the fitting straight line of the gradient direction histogram of the minimum circumscribed circle is obtained as the minimum fitting straight line of each approximate circle area, and the gradient direction histogram of the outer circle adjacent to the minimum circumscribed circle area is taken as the minimum adjacent fitting straight line of each approximate circle area.

[0044] The straight line fitting algorithm selected in one embodiment of the present invention is the Random Sampling Consensus (RANSAC) algorithm, and the implementer can select other straight line fitting algorithms according to actual conditions; among them, the HOG algorithm (Histogram of Oriented Gradients) is a method for feature description in the field of computer vision and image processing, mainly used for target detection and image recognition tasks.

[0045] Furthermore, in order to reflect the degree to which the features of the area near the punch hole in the grayscale image conform to the aperture reflection features, the halo characteristic coefficient of each approximate circular area is calculated based on the average and dispersion of the grayscale values ​​of the pixels of the minimum circumscribed circle and the peripheral circle of the approximate circular area, as well as the absolute value of the slope of the minimum fitting line and the minimum adjacent fitting line. The calculation formula is: ; Wherein, hfc is the halo characteristic coefficient of each approximate circular area; amp min is the mean value of the gradient amplitude of all pixels on the minimum circumscribed circle of each approximate circular area, amp cir is the mean value of the gradient amplitude of all pixels on the outer circle adjacent to each approximate circle area and the minimum circumscribed circle, var gra is the variance of the grayscale values ​​of all pixels on all outer circles of each approximate circular area, k min is the absolute value of the slope of the minimum fitting straight line of each approximate circular area, k cir is the absolute value of the slope of the minimum adjacent fitting straight line of each approximate circular area; To preset the parameter adjustment coefficient, in order to prevent the denominator from being a value of 0, in one embodiment of the present invention The value of is 1, and the implementer can select other values ​​according to the actual situation; sigmoid() is a normalization function.

[0046] It should be noted that for each approximate circular area in the grayscale image of the inner liner plate punching, when the surrounding area of ​​the approximate circular area meets the halo characteristics of reflection, since the color of the area inside the punching is darker, that is, the grayscale value of the pixel points in the area inside the punching is smaller, at this time, if the approximate circular area is the punching area, the minimum circumscribed circle is the edge line of the punching area, the grayscale value of the pixel points is smaller, and the outer circle of the minimum circumscribed circle is the reflective area of ​​the edge of the inner liner plate punching. At this time, the color of the pixel points on the outer circle is lighter, that is, the grayscale value is larger, that is, the difference between the grayscale value of the pixel points on the minimum circumscribed circle and the grayscale value of the pixel points on the adjacent outer circle is larger, and the amp obtained at this time min ,amp cir At the same time, since the boundary line of the halo at the edge of the punching hole is more obvious, that is, there will be a large mutation between the grayscale value of the pixel points on the minimum circumscribed circle and the grayscale value of the pixel points on the adjacent outer circle, at this time, the gradient amplitude and gradient direction of the pixel points on the minimum circumscribed circle and the adjacent outer circle are more evenly distributed (the gradient amplitude is mainly obtained from the difference in the grayscale value of the pixel points from the minimum circumscribed circle to the adjacent outer circle, and the gradient direction is uniformly distributed from 0 degrees to 360 degrees), the obtained gradient direction histogram is smoother, and the obtained minimum fitting straight line and the minimum adjacent fitting straight line are closer to the horizontal line, that is, the slope k of the minimum fitting straight line and the minimum adjacent fitting straight line is min , kcir The smaller the absolute value of , the larger the value of the halo characteristic coefficient hfc obtained at this time; conversely, when the surrounding area of ​​the approximate circular area does not meet the halo characteristics during reflection, the smaller the value of the halo characteristic coefficient hfc obtained at this time.

[0047] It should be noted that in one embodiment of the present invention, the preset halo feature threshold is 0.8, and the implementer may select other values ​​according to the actual situation, which is not specifically limited in the present invention. The approximate circular area whose halo feature coefficient is less than the preset halo feature threshold is regarded as the suspected light spot area, and the area greater than or equal to the halo feature coefficient is regarded as the halo area.

[0048] Furthermore, by analyzing the approximate circular area, the approximate circular area is divided, and the approximate circular area with a halo characteristic coefficient less than the halo characteristic threshold is taken as the suspected light spot area, and the approximate circular area with a halo characteristic coefficient greater than or equal to the halo characteristic threshold is taken as the halo area; In this embodiment, the halo feature coefficient of each approximate circular area is calculated, thereby improving the accuracy of halo feature recognition.

[0049] Step S3: Calculate the spot characteristic coefficient of each suspected spot area, take the suspected spot area whose spot characteristic coefficient is greater than or equal to the spot characteristic threshold as the spot area, and take any area in the halo area and the spot area as the target area; calculate the hole characteristic coefficient of the target area.

[0050] Specifically, for the inner tank plate of the washing machine, when the light spot feature appears on the inner tank plate of the washing machine under normal circumstances, since the edge of the punching has a certain curvature, a light spot will appear in the punching area in the grayscale image of the inner tank plate punching. For the suspected light spot area in the grayscale image of the inner tank plate punching, the grayscale co-occurrence matrix is ​​used to process the suspected light spot area, and the contrast of the suspected light spot area is obtained as the light and dark contrast of each suspected light spot area; at the same time, in order to reflect the dark area in the minimum circumscribed circle of the suspected light spot area and the bright spot in the dark area, the SIFT algorithm is used to process the pixel points in each suspected light spot area, and the feature vector of the minimum circumscribed circle of each suspected light spot area is obtained. Further, the center point of the minimum circumscribed circle of each suspected light spot area is obtained, and a circular area is constructed as the neighborhood of the minimum circumscribed circle of each suspected light spot area with the center point of the minimum circumscribed circle of each suspected light spot area as the center. In one embodiment of the present invention, it is stipulated that the number of the minimum circumscribed circles of the suspected light spot area contained in the circular area is N, and the value of N in one embodiment of the present invention is 10. The implementer can select other values ​​according to actual conditions. Among them, the gray level co-occurrence matrix is ​​a method to describe texture by studying the spatial correlation characteristics of gray levels in an image. The gray level co-occurrence matrix is ​​the probability that the gray level value of a point starting from a pixel with gray level i and leaving a fixed position (with a distance of d and an orientation of θ) is j. All estimated values ​​in the gray level co-occurrence matrix can be expressed in the form of a matrix. The SIFT algorithm (Scale-Invariant Feature Transform) is an algorithm for image feature extraction and matching.

[0051] Furthermore, when the light spot feature appears on the inner plate of the washing machine under normal circumstances, the position of the light spot in each punching hole and the other surrounding punching holes is relatively consistent. Based on the above analysis, the feature vector of the minimum circumscribed circle of each suspected light spot area is used as each feature point, and all feature points in the neighborhood of the minimum circumscribed circle of each suspected light spot area are clustered to obtain each cluster cluster. The clustering algorithm selected in one embodiment of the present invention is the K-means clustering algorithm, and the implementer can select other clustering algorithms according to actual conditions; in order to reflect the consistency of the position of the minimum circumscribed circle light spot in each suspected light spot area, the light spot consistency index of each cluster cluster is constructed, and the calculation formula is: ; Among them, ind fac is the spot consistency coefficient of each cluster; average() is the average function, is a set of silhouette coefficients of all feature points in each cluster. The silhouette coefficient is an indicator used to evaluate clustering quality. It evaluates the clustering effect by quantifying the similarity between the data point and its cluster and other clusters.

[0052] It should be noted that when the minimum circumscribed circle of the suspected light spot area is more consistent with the light spot characteristics of the punching area under normal circumstances, the minimum circumscribed circle of the suspected light spot area is consistent with the position of the light spot in the minimum circumscribed circle of other suspected light spot areas in its neighborhood. At this time, the minimum circumscribed circle of the suspected light spot area has a high similarity with the feature vectors of other suspected light spot areas in its neighborhood, and the clustering effect of the clustering cluster is better, that is, the value of the contour coefficient of all feature points in the clustering cluster is larger, and the light spot consistency coefficient ind obtained at this time is fac The larger the value is, the smaller the obtained spot consistency coefficient ind is. fac The smaller the value of .

[0053] Furthermore, the spot characteristic coefficient of each approximate circular area is obtained based on the spot consistency coefficient and the light-dark contrast, and the calculation formula is: ; Among them, ffc is the spot characteristic coefficient of each suspected spot area; ind fac is the spot consistency coefficient of the cluster where the minimum circumscribed circle of each suspected spot area is located, ldc is the light-dark contrast of the minimum circumscribed circle of each suspected spot area; sigmoid( ) is the normalization function.

[0054] It should be noted that for the minimum circumscribed circle of each suspected light spot area, when the suspected light spot area meets the light spot feature of punching high reflectivity, the light spot consistency coefficient value of the cluster where the minimum circumscribed circle of the suspected light spot area is located is larger, and the light and dark difference between the brighter area and the darker area in the minimum circumscribed circle of the suspected light spot area is larger, that is, the light and dark contrast value is larger, and the value of the obtained light spot characteristic coefficient is larger at this time; otherwise, the value of the obtained light spot characteristic coefficient is smaller. The light spot characteristic coefficients of all suspected light spot areas are obtained, and a preset light spot characteristic threshold is set. In one embodiment of the present invention, the value of the preset light spot characteristic threshold is 0.8, and the implementer can select other values ​​according to actual conditions; for each suspected light spot area, the suspected light spot area whose light spot characteristic coefficient is greater than or equal to the preset light spot characteristic threshold is taken as the light spot area.

[0055] Furthermore, any area in the halo area and the spot area is taken as the target area. For the target area in the grayscale image of the inner liner plate punching, the variance of the grayscale values ​​of all pixels in the target area is calculated. In order to reflect the possibility that the distribution of the grayscale values ​​of the pixels in the target area meets the punching characteristics, the in-hole characteristic coefficient of the target area is obtained based on the discrete degree and average of the grayscale values ​​of the pixels in the target area around the average value. The calculation formula is: ; Among them, pfc is the hole characteristic coefficient of the target area; var ciris the variance of the grayscale values ​​of all pixels in the target area, ave is the mean of the grayscale values ​​of all pixels in the target area, and ave def is the preset grayscale mean, in one embodiment of the present invention, ave def The value of is 5, and implementers can select other values ​​according to actual conditions; To preset the parameter adjustment factor, in order to prevent the denominator from being 0, the value in one embodiment of the present invention is 1, and the implementer can select other values ​​according to actual conditions.

[0056] It should be noted that in the grayscale image of the punching of the liner plate, under normal circumstances, the color of the pixel points of the punching part is darker, the grayscale value of the corresponding pixel points of the punching part is smaller, and the grayscale value of the pixel points in the punching part is more uniform; therefore, based on the above analysis, if the distribution characteristics of the grayscale values ​​of the pixels in the target area meet the distribution characteristics of the grayscale values ​​of the pixels in the punching part, that is, the grayscale value of the pixels in the target area is more uniform, the var cir The smaller the value, the darker the color of the pixels in the target area, that is, the smaller the grayscale value of the pixels in the target area, the smaller the preset grayscale mean value is. The present invention selects a smaller preset grayscale mean value, and reflects the color depth of the grayscale value in the target area through the difference between the average grayscale value of the pixels in the target area and the preset grayscale mean value. When the color of the pixels in the target area is darker, it conforms more to the distribution characteristics of the grayscale value of the pixels in the punching part, and the obtained The value of is smaller, and the value of the hole characteristic coefficient pfc obtained is larger.

[0057] In this embodiment, by obtaining the in-hole characteristic coefficient of the target area, the detection accuracy and efficiency are effectively improved, and the quality control of the punching of the inner liner plate is optimized, thereby improving product reliability and safety.

[0058] Step S4: The target area where the hole feature coefficient is greater than the hole feature threshold is taken as the punching feature area, and the grayscale image of the inner liner plate punching is segmented with each punching feature area as the center, and defect detection is performed on the segmented grayscale image to obtain the punching defect.

[0059] Specifically, for each light spot region and each light ring region, a region where the hole feature coefficient is greater than a preset hole feature threshold is used as a punching feature region.

[0060] For the grayscale image of the punching of the inner liner plate, a mark-based watershed algorithm is used to segment it, and an area with a preset radius centered on each punching feature area is used as a marking area, wherein the value of the preset radius must be smaller than the radius of the punching. In one embodiment of the present invention, the value of the preset radius is 2, and the implementer can select other values ​​according to actual conditions; then, the grayscale image of the punching of the inner liner plate is segmented to obtain a more refined segmented image, and the segmented grayscale image of the punching of the inner liner plate is subjected to defect detection using a template matching algorithm based on an image pyramid to obtain the punching defects of the inner liner plate of the washing machine.

[0061] In this embodiment, the punching feature area is obtained by the hole feature coefficient, the halo feature coefficient and the spot feature coefficient, and the inner tank plate punching grayscale image is marked based on the punching feature area to improve the detection accuracy of the punching defects of the inner tank plate of the washing machine.

Claims

1. A method for detecting punching defects in a washing machine liner, characterized in that: include: Acquire the closed edge line of the grayscale image of the punching hole of the inner liner plate to obtain an approximate circular area in the area surrounded by the closed edge line; Dividing the approximate circular area into a suspected light spot area and a light ring area; Calculate the spot characteristic coefficient of each suspected spot area, and take the suspected spot area whose spot characteristic coefficient is greater than or equal to the spot characteristic threshold as the spot area; the spot characteristic coefficient is positively correlated with the contrast and the spot consistency coefficient, and the spot consistency coefficient represents the consistency of the spot appearance position in each suspected spot area; Taking any area of ​​the halo area and the spot area as the target area; calculating the hole characteristic coefficient of the target area, wherein the hole characteristic coefficient is inversely correlated with the variance and mean of the grayscale values ​​of all pixels in the target area; The target area where the hole characteristic coefficient is greater than the hole characteristic threshold is taken as the punching characteristic area; The punching grayscale image of the liner board is segmented with each punching feature area as the center, and defect detection is performed on the segmented grayscale image to obtain the punching defects.

2. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The process of dividing the approximate circular area includes: taking the approximate circular area whose halo characteristic coefficient is less than the halo characteristic threshold as the suspected light spot area, and taking the approximate circular area whose halo characteristic coefficient is greater than or equal to the halo characteristic threshold as the halo area; the halo characteristic coefficient hfc is: ;Among them, amp min is the mean value of the gradient amplitude of all pixels on the minimum circumscribed circle of each approximate circular area, amp cir is the mean value of the gradient amplitude of all pixels on the outer circle adjacent to each approximate circle area and the minimum circumscribed circle, var gra is the variance of the grayscale values ​​of all pixels on all outer circles of each approximate circular area, k min is the absolute value of the slope of the minimum fitting straight line of each approximate circular area, k cir is the absolute value of the slope of the minimum adjacent fitting straight line of each approximate circular area; is the preset parameter coefficient, and sigmoid() is the normalization function.

3. A method for detecting punching defects in a washing machine liner according to claim 2, characterized in that: The process of obtaining the minimum fitting straight line specifically includes: For the approximate circular area in the grayscale image of the punching hole of the liner plate, the HOG algorithm is used to obtain the gradient direction histogram of all pixel points on the minimum circumscribed circle of the approximate circular area; A straight line fitting process is performed on the values ​​of the intervals in the gradient direction histogram to obtain a minimum fitting straight line.

4. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The hole characteristic coefficient is specifically calculated as: ； Among them, pfc is the hole characteristic coefficient of the target area; var cir is the variance of the grayscale values ​​of all pixels in the target area, ave is the mean of the grayscale values ​​of all pixels in the target area, and ave def is the preset grayscale mean; It is the preset parameter factor.

5. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The light spot characteristic coefficient is: ； Among them, ffc is the spot characteristic coefficient of each suspected spot area; ind fac is the spot consistency coefficient of the cluster where the minimum circumscribed circle of each suspected spot area is located, ldc is the light-dark contrast of the minimum circumscribed circle of each suspected spot area; sigmoid() is the normalization function.

6. A method for detecting punching defects in a washing machine liner according to claim 5, characterized in that: The light spot consistency coefficient is specifically: ； Among them, ind fac is the spot consistency coefficient of the cluster; average() is the average function, It is a set of contour coefficients of all feature points in the clustering cluster, and the clustering cluster is obtained by clustering all feature points in the neighborhood of the minimum circumscribed circle of the suspected light spot area with the feature vector of the minimum circumscribed circle of the suspected light spot area as the feature point, and the feature vector is obtained by processing the pixel points in the minimum circumscribed circle of the suspected light spot area using the SIFT algorithm.

7. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The approximate circular area is an area enclosed by closed edge lines whose circularity is greater than a preset circularity threshold.

8. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The defect acquisition process of the inner liner plate punching specifically includes: The grayscale image of the punching of the inner liner plate is segmented using the watershed algorithm of mark, and the area with a preset radius as the center of each punching feature area is used as the marking area, wherein the value of the preset radius is smaller than the radius of the punching; The grayscale image of the punching of the liner board in the marked area is segmented to obtain a segmented image, and a template matching algorithm based on an image pyramid is used to perform defect detection on the segmented grayscale image of the punching of the liner board to obtain the punching defects of the liner board.

9. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The method for marking the grayscale image of the punching holes of the liner plate comprises: marking the position of the punching feature area on the image using a marking algorithm.

10. A method for detecting punching defects in a washing machine liner according to claim 1, characterized in that: The grayscale image of the punching of the inner liner plate is obtained by collecting the punching image of the inner liner plate using a high-pixel industrial camera, denoising the punching image of the inner liner plate using image denoising, and then grayscale processing the denoised punching image of the inner liner plate, wherein the image denoising is performed using a mean filtering denoising algorithm.

Images