Porcelain surface defect detection method based on image processing

Abstract

The invention discloses a porcelain surface defect detection method based on image processing, and relates to the field of porcelain surface defect detection, and the method comprises the following steps: carrying out bubble state analysis on a porcelain surface bubble image to obtain a porcelain surface bubble abnormal value, carrying out crack state analysis on a porcelain surface crack image to obtain a porcelain surface crack abnormal value, performing porcelain surface deformation state analysis on the porcelain surface deformation image to obtain a porcelain surface deformation value, and evaluating the porcelain surface defect state according to the porcelain surface bubble abnormal value, the porcelain surface crack abnormal value and the porcelain surface deformation value to obtain a porcelain surface defect abnormal value; the porcelain surface defect abnormal value and the set gradient reference interval are compared and analyzed, porcelain surface defect states in different states are correspondingly displayed and alarmed, the appearance quality of porcelain products can be deeply judged, the accuracy of porcelain surface defect judgment can be ensured, and porcelain with different surface qualities can be distinguished in time.

Description

Technical Field

[0001] The invention relates to the field of porcelain surface defect detection, and in particular to a porcelain surface defect detection method based on image processing. Background Art

[0002] As market competition becomes increasingly fierce, the public's requirements for the appearance quality of porcelain products are also increasing. Traditional porcelain surface defect detection is too dependent on manual work, and manual visual inspection is too dependent on the industry level and subjective awareness of the inspectors. Not only is it difficult to meet the requirements in terms of detection speed, but the accuracy of the detection cannot be guaranteed. Therefore, porcelain surface defect detection methods based on image processing have emerged.

[0003] Traditional image processing-based porcelain surface defect detection is difficult to make in-depth judgments on the appearance quality of porcelain products and cannot guarantee the accuracy of porcelain surface defect judgments, resulting in the inability to timely distinguish porcelain with different surface qualities, reducing the overall product quality. Therefore, how to make in-depth and comprehensive judgments on the appearance quality of porcelain products is the main technical problem to be solved by this solution.

[0004] In order to solve the above defects, a technical solution is now provided. Summary of the invention

[0005] In order to solve the technical problems raised by the above background technology, the present invention is proposed. The embodiment of the present invention provides a porcelain surface defect detection method based on image processing.

[0006] The object of the present invention can be achieved by the following technical solution: A porcelain surface defect detection method based on image processing comprises the following steps: Step 1: Analyze the bubble state of the bubble image on the porcelain surface to obtain the bubble abnormality value on the porcelain surface; Step 2: Analyze the crack state of the porcelain surface image crack image to obtain the porcelain surface crack shape anomaly value; Step 3: Analyze the deformation state of the porcelain surface on the porcelain surface deformation image to obtain the porcelain surface deformation value; Step 4: Evaluate the state of the porcelain surface defects by using the porcelain surface bubble anomaly value, porcelain surface crack anomaly value and porcelain surface deformation value to obtain the porcelain surface defect anomaly value; The bubble anomaly value, crack anomaly value and deformation value of the porcelain surface are normalized, and a parallelogram is constructed with the length of the bubble anomaly value and crack anomaly value as the side length and height, and a circle is constructed with the length of the deformation value as the radius, wherein the center of the circle coincides with the upper right vertex of the parallelogram, and the formed area is identified and marked as the porcelain surface defect anomaly value; Step 5: Compare and analyze the abnormal value of the porcelain surface defect with the set gradient reference interval, and display and alarm the porcelain surface defect status in different states accordingly.

[0007] Furthermore, the specific analysis steps of the bubble abnormality value on the porcelain surface are as follows: Step 107: Compare the irregularity value of each bubble defect area with the set reference irregularity value threshold. When the irregularity value of each bubble defect area is greater than the set reference irregularity value threshold, the corresponding bubble defect area is a severely irregular area. When the irregularity value of each bubble defect area is equal to the set reference irregularity value threshold, the corresponding bubble defect area is a moderately irregular area. When the irregularity value of each bubble defect area is less than the set reference irregularity value threshold, the corresponding bubble defect area is a slightly irregular area. The proportions of slightly irregular areas, moderately irregular areas and severely irregular areas are counted to obtain the individual values ​​of the bubble defect areas, and the total area of ​​the bubble defect areas is counted. The formula is used to calculate the bubble abnormality value cqz of the porcelain surface.

[0008] Furthermore, the specific analysis steps of the irregularity value of each bubble defect area are as follows: Step 105: The boundary pixel coordinates of the bubble defect area are obtained by edge detection algorithm, which are (a1, b1), (a2, b2), ..., (am, bm), where m is the boundary pixel number. The boundary pixel coordinates form a closed curve. The parameter c is set to a value between 0 and 1. As c increases from 0 to 1, (a(c), b(c)) passes through all pixel coordinates of the defect area boundary in turn. M boundary points are set. The pixel coordinates after parameter processing are (a(c m ), b (c m )), c m is the parameter value; Step 106: Through discrete Fourier transform and formula calculation, the Fourier coefficient F(d) is obtained, and F(d) is expanded into real and imaginary parts to obtain the real part A(d) of F(d) and the imaginary part B(d) of F(d). The amplitude of F(d) |F(d)| is obtained by formula analysis, and d=0 to d=M / 3 is divided into a low frequency band, M / 2>d>M / 3 is divided into a medium frequency band, and d≥M / 2 is defined as a high frequency band. The irregular value bg of each bubble defect area is obtained by analysis.

[0009] Furthermore, the specific analysis steps of the bubble defect area are as follows: Step 101: Arrange multiple groups of cameras in the porcelain production area to obtain images of the porcelain surface, and perform Gaussian filtering on the images, specifically, perform weighted average calculation on each pixel in the image and its adjacent domain pixels, where the weight is determined by the Gaussian function, so as to smooth the image, reduce the noise of the image, and stretch the contrast of the image after noise processing by setting a reference grayscale range; Step 102: matching the processed porcelain surface image with the porcelain bubble image in the data to obtain the gray threshold range of the bubble; Step 103: taking the lower left corner of the image as the coordinate origin, horizontal rightward as the positive direction of the x-axis, and vertical upward as the positive direction of the y-axis, setting the grayscale threshold range of the bubble, determining a coordinate point within the range in the porcelain surface image, and defining it as a starting seed point with coordinates (x1, y1), selecting eight coordinate points, namely, (x1, y1+1), (x1+1, y1), (x1, y1-1), (x1-1, y1), (x1-1, y1-1), (x1+1, y1+1), (x1-1, y1+1) and (x1+1, y1-1), and determining whether the pixel grayscale value of each coordinate point is within the grayscale threshold range of the bubble; if it is within the set grayscale threshold range of the bubble, merging the pixel point into the defined growth area; Step 104: The seed point is judged from left to right and from bottom to top for each pixel point of the porcelain surface image, and step 103 is repeated until the seed point traverses to the upper right corner of the porcelain surface image, and the growth area no longer expands, forming several bubble defect areas for the final segmentation.

[0010] Furthermore, the specific analysis steps of the crack shape abnormal value of the porcelain surface are as follows: Step 206: Calculate the porcelain surface crack shape anomaly value lzz by using the formula of the porcelain surface crack curvature value, the porcelain surface crack length density value, the porcelain surface number density value and the porcelain surface crack width value.

[0011] Furthermore, the specific analysis steps of the porcelain surface crack curvature value, the porcelain surface crack length density value, the porcelain surface number density value and the porcelain surface crack width value are as follows: Step 205: Obtain the slope of the crack segment and another adjacent crack segment, divided into k1 and k2, and the distance xj between the two adjacent crack segments. According to the set formula ql=δ×|k1-k2| / xj, obtain the curvature value ql of the crack segment, δ is the set correction factor coefficient, and process the average value of the curvature value of each crack segment to obtain the porcelain surface crack curvature value ql, sum and count the Euclidean length values ​​of each crack to obtain the total Euclidean length of the crack, obtain the porcelain surface area, the ratio of the total Euclidean length of the crack to the porcelain surface area, marked as the porcelain surface crack length density value, calculate the ratio of the total number of cracks to the porcelain surface area, marked as the porcelain surface number density value sm, and process the average value of the width of each crack to obtain the porcelain surface crack width value lk.

[0012] Furthermore, the specific analysis steps of the Euclidean length value of each crack and the width value of each crack are as follows: Step 201: applying an edge detection operator to the porcelain surface image after the stretching noise processing to obtain the edge area where the crack exists; Step 202: The crack line in the porcelain surface image is expressed in the form of polar coordinates (ρ, θ), where ρ refers to the vertical distance from the origin to the line, and θ refers to the angle between the line segment of the vertical distance and the positive direction of the x-axis, and is discretized at a certain angle value, and the value range of ρ is determined according to the size of the porcelain surface image, and discretized at a certain distance value to obtain a curve of the edge point; Step 203: Substitute each edge pixel point in the edge region where the crack exists into step 202 for processing to obtain the curve graph of each edge point. There is an intersection between the curve graphs of each edge point. When the density of the intersection point is greater than the set reference value, the point (ρ h ,θ h ) is the parameter of the crack line, h is the number of the crack line, and the crack line equation is obtained; Step 204: Annotate the determined crack line equation on the porcelain surface image, obtain the coordinates of the two endpoints of each crack (xp1, yp1) and (xp2, yp2), calculate the Euclidean length value od of each crack by the formula, and from the starting end point of each crack, gradually obtain the gray value of each point at a certain distance on both sides of the normal direction of the crack line, obtain the gray value on each crack line, and make a difference with the gray value gradually obtained at a certain distance on both sides of the normal direction. When the gray difference is less than a preset threshold, stop obtaining, and the range obtained on both sides is the width value of the crack coordinate point. Repeat the above steps from the starting end to the end end of the crack to obtain the width value of each crack coordinate point, and divide the sum of each width value by the length value of the crack to obtain the width value of each crack.

[0013] Furthermore, the specific steps of analyzing the surface deformation value of the porcelain are as follows: Step 301: Obtain the size of the porcelain surface image marked as Z×G, where Z is the height pixel value of the image, G is the width pixel value of the image, set the pixel coordinates in the image to (z, g), and calculate the total coordinates (z tot , g tot ), the total coordinates (z cent , g cent ), the formula is calculated to obtain the porcelain surface displacement value zyz; Step 302: Obtain the porcelain surface image and obtain the outer contour of the porcelain bottle mouth through the edge detection algorithm, obtain the total pixel value T of the porcelain surface image, and the coordinates of each contour point are (x k ,y k ), where k is the number of the outer contour point. According to the set formula, the center of the porcelain bottle mouth (x zx ,y zx ), establish the horizontal symmetry axis y=y through the center coordinate point of the porcelain bottle mouth zx and the vertical axis of symmetry x=x zx , analyzing to obtain regional pixel value difference 1 and regional pixel value difference 2, processing and analyzing the regional pixel value difference 1 and the regional pixel value difference 2 to obtain the regional pixel value total difference, processing and analyzing the regional pixel value total difference to obtain the symmetry deviation of the porcelain bottle mouth; Step 303: perform weighted calculation on the porcelain surface displacement value and the porcelain bottle mouth symmetry deviation value, and multiply them by the corresponding correction factor coefficient to obtain the porcelain surface deformation value.

[0014] Furthermore, the specific analysis steps for displaying and alarming the surface defects of porcelain in different states are as follows: When the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA1, a corresponding surface defect severe alarm signal is issued, and label one is displayed on the display screen; when the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA2, a corresponding surface defect moderate alarm signal is issued, and label two is displayed on the display screen; when the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA3, label three is displayed on the display screen.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention obtains the porcelain surface bubble anomaly value by analyzing the bubble state of the porcelain surface bubble image, obtains the porcelain surface crack anomaly value by analyzing the crack state of the porcelain surface crack image, obtains the porcelain surface crack anomaly value by analyzing the porcelain surface deformation state of the porcelain surface deformation image, and obtains the porcelain surface deformation value. It can conduct in-depth analysis on the surface bubbles, surface cracks and surface deformation of the porcelain, thereby ensuring the accuracy of porcelain surface defect judgment.

[0016] 2. The present invention evaluates the state of porcelain surface defects by measuring the bubble anomaly value, crack anomaly value and deformation value of the porcelain surface to obtain the porcelain surface defect anomaly value, compares and analyzes the porcelain surface defect anomaly value with the set gradient reference interval, and displays and alarms the porcelain surface defect states in different states accordingly, which can distinguish porcelain with different surface qualities in time, improve the overall product quality, improve the production efficiency of the enterprise, and reduce the production cost of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. The following drawings are not intentionally scaled according to the actual sizes, and the focus is on illustrating the main purpose of the present invention.

[0018] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present invention.

[0020] like Figure 1 As shown, the porcelain surface defect detection method based on image processing includes: Step 1: Obtain a bubble image on the porcelain surface, analyze the bubble state of the porcelain surface image, and obtain the bubble abnormality value on the porcelain surface; Step 2: Obtain a crack image of the porcelain surface image, and analyze the crack state of the porcelain surface image based on the image to obtain a crack shape anomaly value of the porcelain surface; Step 3: Obtain a deformation image of the porcelain surface, analyze the deformation state of the porcelain surface, and obtain a deformation value of the porcelain surface; Step 4: receiving the porcelain surface bubble abnormality value, the porcelain surface crack abnormality value and the porcelain surface deformation value, and evaluating the state of the porcelain surface defect based on the value to obtain the porcelain surface defect abnormality value; Step 5: Compare and analyze the abnormal value of the porcelain surface defect with the set gradient reference interval, and display and alarm the porcelain surface defect status in different states accordingly.

[0021] The process of obtaining the bubble state analysis on the porcelain surface specifically includes the following steps: Step 101: multiple groups of cameras are arranged in the porcelain production area to obtain images of the porcelain surface, and the images are subjected to Gaussian filtering, specifically, weighted average calculation is performed on each pixel in the image and its adjacent domain pixels, the weight is determined by a Gaussian function, and the closer the pixel is to the center pixel, the greater the weight is, so as to smooth the image and reduce the noise of the image, and the contrast of the image after noise processing is stretched by setting a reference grayscale range, so that bubbles are easier to be identified in the porcelain surface image; Step 102: matching the processed porcelain surface image with the porcelain bubble image in the data to obtain the gray threshold range of the bubble; Step 103: taking the lower left corner of the image as the coordinate origin, horizontal rightward as the positive direction of the x-axis, and vertical upward as the positive direction of the y-axis, setting the grayscale threshold range of the bubble, determining a coordinate point within the range in the porcelain surface image, and defining it as a starting seed point with coordinates (x1, y1), selecting eight coordinate points, namely, (x1, y1+1), (x1+1, y1), (x1, y1-1), (x1-1, y1), (x1-1, y1-1), (x1+1, y1+1), (x1-1, y1+1) and (x1+1, y1-1), and determining whether the pixel grayscale value of each coordinate point is within the grayscale threshold range of the bubble; if it is within the set grayscale threshold range of the bubble, merging the pixel point into the defined growth area; Step 104: the seed point is sequentially determined from left to right and from bottom to top for each pixel point of the porcelain surface image, and step 103 is continuously repeated until the seed point traverses to the upper right corner of the porcelain surface image, and the growth area is no longer expanded, thus forming a number of bubble defect areas for the final segmentation; Step 105: Obtain the boundary pixel coordinates of the bubble defect area through the edge detection algorithm, which are (a1, b1), (a2, b2), ..., (am, bm), where m is the boundary pixel number. The boundary pixel coordinates form a closed curve. The parameter c is set to a value between 0 and 1. When c=0, (a(0), b(0)) represents the starting point on the boundary. When c=1, (a(1), b(1)) represents the end point on the boundary. As c increases from 0 to 1, (a(c), b(c)) passes through all pixel coordinate points on the boundary of the defect area in turn. M boundary points are set. The pixel coordinate points after parameter processing are (a(c m ), b (c m )), where m=0, 1, 2, ..., M-1, c m is the parameter value; Step 106: Through discrete Fourier transform, according to the formula Get the Fourier coefficient F(d), where i is the imaginary unit, and pass F(d) through e -iθ=cosθ-isinθ, expand it into real and imaginary parts, and get the real part A(d) of F(d) and the imaginary part B(d) of F(d). According to the formula , get the amplitude of F(d) |F(d)|, divide d=0 to d=M / 3 into the low frequency band, M / 2>d>M / 3 into the medium frequency band, and d≥M / 2 into the high frequency band. According to the set formula , the irregular value bg of each bubble defect area is obtained, where M01, M02 and MO3 are the set weight factor coefficients of the high frequency band amplitude sum, the medium frequency band amplitude sum, and the low frequency band amplitude sum, all of which are positive integers. The weight factor coefficient is used to reduce the deviation in the formula calculation process, so as to calculate more accurate parameter data; Step 107: Compare the irregularity value of each bubble defect area with the set reference irregularity value threshold. When the irregularity value of each bubble defect area is greater than the set reference irregularity value threshold, the corresponding bubble defect area is a severe irregularity area. When the irregularity value of each bubble defect area is equal to the set reference irregularity value threshold, the corresponding bubble defect area is a moderate irregularity area. When the irregularity value of each bubble defect area is less than the set reference irregularity value threshold, the corresponding bubble defect area is a mild irregularity area. The number of mild irregularity areas, moderate irregularity areas and severe irregularity areas is counted. The proportion of irregular areas is marked as qb, zb and yb respectively, and the individual values ​​of bubble defect areas are obtained, and the total area of ​​the bubble defect areas are counted, marked as qg and qz respectively, and normalized with the proportion of mild irregular areas, moderate irregular areas and severe irregular areas. According to the set formula cqz=f1×yb+f2×zb+f3 / qb+f4×qg+f5×qz, the bubble abnormality value cqz of the porcelain surface is calculated, where f1, f2, f3, f4 and f5 are set reference correction factor coefficients, and the specific values ​​are determined by personnel in this professional field.

[0022] The process of obtaining the surface crack state analysis of porcelain specifically includes the following steps: Step 201: applying an edge detection operator to the porcelain surface image after the stretching noise processing to obtain the edge area where the crack exists; Step 202: The crack line in the porcelain surface image is expressed in the form of polar coordinates (ρ, θ), where ρ refers to the vertical distance from the origin to the line, and θ refers to the angle between the line segment of the vertical distance and the positive direction of the x-axis. The value range of θ is set to [-π / 2, π / 2], and discretized at a certain angle value, and the specific certain angle value can be 0.5°. The value range of ρ is determined according to the size of the porcelain surface image, and discretized at a certain distance value, where the certain distance value is 1 pixel unit, to obtain the curve of the edge point; Step 203: Substitute each edge pixel point in the edge region where the crack exists into step 202 for processing to obtain the curve graph of each edge point. There is an intersection between the curve graphs of each edge point. When the density of the intersection point is greater than the set reference value, the point (ρ h ,θ h ) is the parameter of the crack line, h is the number of the crack line, and the crack line equation is y=-cosθ h ×x / sinθ h +ρ h / sinθ h ; Step 204: Annotate the determined crack line equation on the porcelain surface image, obtain the coordinates of the two endpoints of each crack (xp1, yp1) and (xp2, yp2), and calculate the value of the equation according to the formula , calculate and obtain the Euclidean length value od of each crack, starting from the starting end point of each crack, and gradually obtain the grayscale value of each point at a certain distance on both sides of the crack line along the normal direction, where the certain distance is 1 pixel unit, obtain the grayscale value on each crack line, and make a difference with the grayscale value gradually obtained at a certain distance on both sides of the normal direction. When the grayscale difference is less than a preset threshold, stop obtaining, and the range obtained on both sides is the width value of the crack coordinate point. Repeat the above steps from the starting end to the end end of the crack to obtain the width value of each crack coordinate point, and divide the sum of each width value by the length value of the crack to obtain the width value of each crack; Step 205: Obtain the slope of the crack segment and another adjacent crack segment, which are divided into k1 and k2, and the distance xj between the two adjacent crack segments. According to the set formula ql=δ×|k1-k2| / xj, obtain the curvature value ql of the crack segment, where δ is the set correction factor coefficient used to improve the accuracy of the calculation. The curvature values ​​of each crack segment are averaged to obtain the porcelain surface crack curvature value ql. The Euclidean length values ​​of each crack are summed and counted to obtain the total Euclidean length of the crack, and the porcelain surface area is obtained. The ratio of the total Euclidean length of the crack to the porcelain surface area is marked as the porcelain surface crack length density value cm. The ratio of the total number of cracks to the porcelain surface area is calculated and marked as the porcelain surface number density value sm. The width values ​​of each crack are averaged to obtain the porcelain surface crack width value lk. Step 206: Normalize the porcelain surface crack curvature value ql, the porcelain surface crack length density value cm, the porcelain surface number density value sm and the porcelain surface crack width value lk and substitute them into the set formula: The calculation is performed to obtain the porcelain surface crack shape anomaly value lzz, where LX1, LX2, LX3 and LX4 are preset weight factors of the porcelain surface crack curvature value, the porcelain surface crack length density value, the porcelain surface number density value and the porcelain surface crack width value, respectively. is the preset correction factor, the specific value is 0.841, and e is a natural constant, the value is 2.718.

[0023] The process of obtaining the porcelain surface deformation state analysis specifically includes the following steps: Step 301: Obtain a porcelain surface image with dimensions Z×G, where Z is the image height pixel value, G is the image width pixel value, set the pixel coordinates in the image to (z, g), the value range of z is [0, Z-1], the value range of g is [0, G-1], according to the formula , calculate the total coordinates (z tot , g tot ), the total pixel value of the porcelain surface image is ZZ=Z×G, according to the formula , calculate the total coordinates (z cent , g cent ), according to the formula , get the porcelain surface displacement value zyz, where p1 and p2 are the set correction factor coefficients, A1 is the central horizontal coordinate of the standard template image in the database, and A2 is the central vertical coordinate of the standard template image in the database; Step 302: Obtain the porcelain surface image and obtain the outer contour of the porcelain bottle mouth through the edge detection algorithm, obtain the total pixel value T of the porcelain surface image, and the coordinates of each contour point are (x k ,y k ), where k is the number of the outer contour point, k=1, 2, 3, ..., T, according to the set formula , calculate the center of the porcelain bottle mouth (x zx ,y zx ), establish the horizontal symmetry axis y=y through the center coordinate point of the porcelain bottle mouth zx and the vertical axis of symmetry x=x zx , when x>x zx And y>y zx , the area corresponds to the upper right area, when x>x zx And y <y zx , the area corresponds to the lower right area, when x <x zx And y>y zx , the area corresponds to the upper left area, when x <x zx And y <y zx, the area corresponds to the lower left area, obtain the pixel values ​​of the upper right area, the lower right area, the upper left area and the lower left area, subtract the pixel value of the upper right area from the pixel value of the upper left area to obtain the regional pixel value difference one, subtract the pixel value of the lower right area from the pixel value of the lower left area to obtain the regional pixel value difference two, take the absolute value of the regional pixel value difference one and the regional pixel value difference two respectively and then sum them up to obtain the regional pixel value total difference, take the difference between the regional pixel value total difference and the set reference pixel value total difference and take the absolute value to obtain the symmetry deviation of the porcelain bottle mouth; Step 303: perform weighted calculation on the porcelain surface displacement value and the porcelain bottle mouth symmetry deviation value, and multiply them by the corresponding correction factor coefficient to obtain the porcelain surface deformation value.

[0024] Among them, the acquisition process of porcelain surface defect evaluation status analysis specifically includes the following steps: The porcelain surface bubble anomaly value cqz, the porcelain surface crack anomaly value lzz and the porcelain surface deformation value are normalized, and a parallelogram is constructed with the length of the porcelain surface bubble anomaly value cqz and the porcelain surface crack anomaly value lzz as the side length and height, and a circle is constructed with the length of the porcelain surface deformation value as the radius, in which the center of the circle coincides with the upper right vertex of the parallelogram, and the non-overlapping area is identified and marked as the porcelain surface defect anomaly value.

[0025] The acquisition process of porcelain surface defect status evaluation and judgment specifically includes the following steps: When the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA1, a serious surface defect alarm is issued, and the display screen set up in the factory production line displays "the current porcelain surface defect is serious"; When the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA2, a corresponding surface defect moderate alarm is issued, and the display screen set on the factory production line displays "Current porcelain surface defect moderate"; When the surface defect anomaly value of the porcelain is within the set defect anomaly value gradient reference interval AA3, no alarm is issued accordingly, and the display screen set up on the factory production line displays "The current surface defect of the porcelain is slight", where the gradient reference interval AA1>AA2>AA3.

[0026] The above is an explanation of the present invention and should not be considered as a limitation thereof. Although several exemplary embodiments of the present invention have been described, it will be readily appreciated by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Therefore, all such modifications are intended to be included within the scope of the present invention as defined in the claims. It should be understood that the above is an explanation of the present invention and should not be considered as being limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A porcelain surface defect detection method based on image processing, characterized in that: The following steps are involved: Step 1: Analyze the bubble state of the bubble image on the porcelain surface to obtain the bubble abnormality value on the porcelain surface; Step 2: Analyze the crack state of the porcelain surface image crack image to obtain the porcelain surface crack shape anomaly value; Step 3: Analyze the deformation state of the porcelain surface on the porcelain surface deformation image to obtain the porcelain surface deformation value; Step 4: Evaluate the state of the porcelain surface defects by using the porcelain surface bubble anomaly value, porcelain surface crack anomaly value and porcelain surface deformation value to obtain the porcelain surface defect anomaly value; The bubble anomaly value, crack anomaly value and deformation value of the porcelain surface are normalized, and a parallelogram is constructed with the length of the bubble anomaly value and crack anomaly value as the side length and height, and a circle is constructed with the length of the deformation value as the radius, wherein the center of the circle coincides with the upper right vertex of the parallelogram, and the formed area is identified and marked as the porcelain surface defect anomaly value; Step 5: Compare and analyze the abnormal value of the porcelain surface defect with the set gradient reference interval, and display and alarm the porcelain surface defect status in different states accordingly.

2. The method for detecting surface defects of porcelain based on image processing according to claim 1, characterized in that: The specific analysis steps of the bubble abnormality value on the porcelain surface are as follows: Step 107: Compare the irregularity value of each bubble defect area with the set reference irregularity value threshold. When the irregularity value of each bubble defect area is greater than the set reference irregularity value threshold, the corresponding bubble defect area is a severely irregular area. When the irregularity value of each bubble defect area is equal to the set reference irregularity value threshold, the corresponding bubble defect area is a moderately irregular area. When the irregularity value of each bubble defect area is less than the set reference irregularity value threshold, the corresponding bubble defect area is a slightly irregular area. The proportions of slightly irregular areas, moderately irregular areas and severely irregular areas are counted to obtain the individual values ​​of the bubble defect areas, and the total area of ​​the bubble defect areas is counted. The formula is used to calculate the bubble abnormality value cqz of the porcelain surface.

3. The porcelain surface defect detection method based on image processing according to claim 2 is characterized in that: The specific analysis steps of the irregularity values ​​of each bubble defect area are as follows: Step 105: The boundary pixel coordinates of the bubble defect area are obtained by edge detection algorithm, which are (a1, b1), (a2, b2), ..., (am, bm), where m is the boundary pixel number. The boundary pixel coordinates form a closed curve. The parameter c is set to a value between 0 and 1. As c increases from 0 to 1, (a(c), b(c)) passes through all pixel coordinates of the defect area boundary in turn. M boundary points are set. The pixel coordinates after parameter processing are (a(c m ), b (c m )), c m is the parameter value; Step 106: Through discrete Fourier transform and formula calculation, the Fourier coefficient F(d) is obtained, and F(d) is expanded into real and imaginary parts to obtain the real part A(d) of F(d) and the imaginary part B(d) of F(d). The amplitude of F(d) |F(d)| is obtained by formula analysis, and d=0 to d=M / 3 is divided into a low frequency band, M / 2>d>M / 3 is divided into a medium frequency band, and d≥M / 2 is defined as a high frequency band. The irregular value bg of each bubble defect area is obtained by analysis.

4. The method for detecting surface defects of porcelain based on image processing according to claim 3, characterized in that: The specific analysis steps of the bubble defect area are as follows: Step 101: Arrange multiple groups of cameras in the porcelain production area to obtain images of the porcelain surface, and perform Gaussian filtering on the images, specifically, perform weighted average calculation on each pixel in the image and its adjacent domain pixels, where the weight is determined by the Gaussian function, so as to smooth the image, reduce the noise of the image, and stretch the contrast of the image after noise processing by setting a reference grayscale range; Step 102: matching the processed porcelain surface image with the porcelain bubble image in the data to obtain the gray threshold range of the bubble; Step 103: taking the lower left corner of the image as the coordinate origin, horizontal rightward as the positive direction of the x-axis, and vertical upward as the positive direction of the y-axis, setting the grayscale threshold range of the bubble, determining a coordinate point within the range in the porcelain surface image, and defining it as a starting seed point with coordinates (x1, y1), selecting eight coordinate points, namely, (x1, y1+1), (x1+1, y1), (x1, y1-1), (x1-1, y1), (x1-1, y1-1), (x1+1, y1+1), (x1-1, y1+1) and (x1+1, y1-1), and determining whether the pixel grayscale value of each coordinate point is within the grayscale threshold range of the bubble; if it is within the set grayscale threshold range of the bubble, merging the pixel point into the defined growth area; Step 104: The seed point is judged from left to right and from bottom to top for each pixel point of the porcelain surface image, and step 103 is repeated until the seed point traverses to the upper right corner of the porcelain surface image, and the growth area no longer expands, forming several bubble defect areas for the final segmentation.

5. The porcelain surface defect detection method based on image processing according to claim 1 is characterized in that: The specific analysis steps of the abnormal value of the crack shape on the porcelain surface are as follows: Step 206: Calculate the porcelain surface crack curvature value, the porcelain surface crack length density value, the porcelain surface number density value and the porcelain surface crack width value to obtain the porcelain surface crack shape anomaly value lzz.

6. The method for detecting surface defects of porcelain based on image processing according to claim 5, characterized in that: The specific analysis steps of the porcelain surface crack curvature value, the porcelain surface crack length density value, the porcelain surface number density value and the porcelain surface crack width value are as follows: Step 205: Obtain the slope of the crack segment and another adjacent crack segment, divided into k1 and k2, and the distance xj between the two adjacent crack segments. According to the set formula ql=δ×|k1-k2| / xj, obtain the curvature value ql of the crack segment, δ is the set correction factor coefficient, and process the average value of the curvature value of each crack segment to obtain the porcelain surface crack curvature value ql, sum and count the Euclidean length values ​​of each crack to obtain the total Euclidean length of the crack, obtain the porcelain surface area, the ratio of the total Euclidean length of the crack to the porcelain surface area, marked as the porcelain surface crack length density value, calculate the ratio of the total number of cracks to the porcelain surface area, marked as the porcelain surface number density value sm, and process the average value of the width of each crack to obtain the porcelain surface crack width value lk.

7. The method for detecting porcelain surface defects based on image processing according to claim 6, characterized in that: The specific analysis steps of the Euclidean length value of each crack and the width value of each crack are as follows: Step 201: applying an edge detection operator to the porcelain surface image after the stretching noise processing to obtain the edge area where the crack exists; Step 202: The crack line in the porcelain surface image is expressed in the form of polar coordinates (ρ, θ), where ρ refers to the vertical distance from the origin to the line, and θ refers to the angle between the line segment of the vertical distance and the positive direction of the x-axis, and is discretized at a certain angle value, and the value range of ρ is determined according to the size of the porcelain surface image, and discretized at a certain distance value to obtain a curve of the edge point; Step 203: Substitute each edge pixel point in the edge region where the crack exists into step 202 for processing to obtain the curve graph of each edge point. There is an intersection between the curve graphs of each edge point. When the density of the intersection point is greater than the set reference value, the point (ρ h ,θ h ) is the parameter of the crack line, h is the number of the crack line, and the crack line equation is obtained; Step 204: Annotate the determined crack line equation on the porcelain surface image, obtain the coordinates of the two endpoints of each crack (xp1, yp1) and (xp2, yp2), calculate the Euclidean length value od of each crack by the formula, and from the starting end point of each crack, gradually obtain the gray value of each point at a certain distance on both sides of the normal direction of the crack line, obtain the gray value on each crack line, and make a difference with the gray value gradually obtained at a certain distance on both sides of the normal direction. When the gray difference is less than a preset threshold, stop obtaining, and the range obtained on both sides is the width value of the crack coordinate point. Repeat the above steps from the starting end to the end end of the crack to obtain the width value of each crack coordinate point, and divide the sum of each width value by the length value of the crack to obtain the width value of each crack.

8. The method for detecting surface defects of porcelain based on image processing according to claim 1, characterized in that: The specific analysis steps of the porcelain surface deformation value are as follows: Step 301: Obtain the size of the porcelain surface image marked as Z×G, where Z is the height pixel value of the image, G is the width pixel value of the image, set the pixel coordinates in the image to (z, g), and calculate the total coordinates (z tot , g tot ), the total coordinates (z cent , g cent ), the formula is calculated to obtain the porcelain surface displacement value zyz; Step 302: Obtain the porcelain surface image and obtain the outer contour of the porcelain bottle mouth through the edge detection algorithm, obtain the total pixel value T of the porcelain surface image, and the coordinates of each contour point are (x k ,y k ), where k is the number of the outer contour point. According to the set formula, the center of the porcelain bottle mouth (x zx ,y zx ), establish the horizontal symmetry axis y=y through the center coordinate point of the porcelain bottle mouth zx and the vertical axis of symmetry x=x zx , analyzing to obtain regional pixel value difference 1 and regional pixel value difference 2, processing and analyzing the regional pixel value difference 1 and the regional pixel value difference 2 to obtain the regional pixel value total difference, processing and analyzing the regional pixel value total difference to obtain the symmetry deviation of the porcelain bottle mouth; Step 303: perform weighted calculation on the porcelain surface displacement value and the porcelain bottle mouth symmetry deviation value, and multiply them by the corresponding correction factor coefficient to obtain the porcelain surface deformation value.

9. The method for detecting surface defects of porcelain based on image processing according to claim 1, characterized in that: The specific analysis steps for displaying and alarming the different states of porcelain surface defects are as follows: When the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA1, a corresponding surface defect severe alarm signal is issued, and label one is displayed on the display screen; when the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA2, a corresponding surface defect moderate alarm signal is issued, and label two is displayed on the display screen; when the surface defect abnormality value of the porcelain is within the set defect abnormality value gradient reference interval AA3, label three is displayed on the display screen.

Images