Quality inspection method for precision metal foil part stamping processing

Abstract

The present application relates to the technical field of metal foil wrinkle defect detection, and particularly relates to a quality inspection method for precision metal foil part stamping processing. The method acquires an image of the metal foil; based on the distribution of the gray value in each local area in the image, the wrinkle degree of each local area is acquired; according to the gray distribution difference between the adjacent rows and the adjacent columns in each local area, the reflection degree of each local area is acquired and the gray value of each pixel point in each local area is corrected; according to the wrinkle degree and the reflection degree of each local area before and after correction, the wrinkle prominence degree of each local area is acquired, and the quality of the metal foil part stamping processing is judged. The present application acquires the wrinkle prominence degree, accurately determines whether the corresponding local area has a wrinkle, effectively improves the accuracy of the wrinkle defect recognition in the metal foil, and is beneficial to more accurately judging the quality of the metal foil part stamping processing.

Description

Technical Field

[0001] This invention relates to the field of metal foil wrinkle defect detection technology, and more specifically to a quality inspection method for the stamping process of precision metal foil parts. Background Technology

[0002] Metal foil, as an ultra-thin metal sheet processed through precision techniques such as rolling and electrolysis, occupies an irreplaceable core position in fields with stringent requirements for product size, weight, and performance, including precision instruments, aerospace, and electronic devices, due to its extremely high dimensional accuracy, surface finish, and material uniformity. The surface quality of metal foil parts is a key factor determining the performance and reliability of the final product. However, various microscopic surface defects are easily generated during the stamping process of metal foil, such as indentations or scratches caused by stamping, burrs at edges and corners, and localized wrinkles caused by uneven stress or heat distribution. These defects can directly damage the structural integrity of the parts, affect subsequent assembly accuracy, and even lead to product malfunction. Therefore, comprehensive and accurate quality inspection of stamped metal foil is an indispensable and crucial step in the production process.

[0003] Existing methods use machine vision to detect surface defects in the stamping process of precision metal foil parts. However, they face a major challenge: the reflection phenomenon on the foil surface seriously interferes with defect identification. Unlike other thicker metal materials, the surface of stamped metal foil exhibits extremely high flatness. This characteristic makes it difficult for incident light to form diffuse reflection in multiple directions. Instead, it is reflected in a single direction in a highly concentrated specular reflection form, thus forming a large area of ​​bright light in the detection image. For defects such as cracks or scratches on the surface of metal foil that damage the integrity of the foil, color differences can be used for identification. However, wrinkle defects are essentially non-uniform plastic deformations that occur locally in the foil, manifesting only as changes in shape such as protrusions and depressions on the surface. The material and color remain completely consistent with the normal area. Therefore, they can only be identified by the slight grayscale changes caused by differences in light reflection. However, the ultra-thin nature of metal foil makes it extremely sensitive to light, and the reflection phenomenon is much more significant than that of other foil materials. Moreover, the grayscale fluctuations caused by reflection are highly similar to the grayscale characteristics of wrinkle defects. As a result, existing methods cannot accurately identify wrinkle defects, leading to a seriously low accuracy rate in defect detection of metal foil, which makes it difficult to meet the high-quality inspection requirements of metal foil parts in the precision manufacturing field. Summary of the Invention

[0004] To address the technical problem of inaccurate identification of wrinkles and defects after stamping of metal foil, the present invention aims to provide a quality inspection method for stamping precision metal foil parts. The specific technical solution adopted is as follows:

[0005] This invention provides a quality inspection method for the stamping process of precision metal foil parts, the method comprising the following steps:

[0006] Acquire images of metal foil;

[0007] The image is divided into local regions, and the degree of wrinkling in each local region is obtained based on the distribution of gray values ​​in each local region.

[0008] Based on the difference in grayscale distribution between adjacent rows and columns in each local region, the reflectivity of each local region is obtained and the grayscale value of the pixels in each local region is corrected;

[0009] Based on the degree of wrinkling and reflectivity of each local area before correction, the significance of wrinkles in each local area is obtained.

[0010] The quality of stamping of metal foil parts is judged based on the degree of wrinkle.

[0011] Furthermore, the method for obtaining the degree of wrinkles is as follows:

[0012] For any local region, the gray level corresponding to the largest number in the gray-level histogram of that local region is taken as the target gray level;

[0013] Based on the positional distribution of the target gray level in the gray-level histogram, the degree of concentration of the gray-level histogram is obtained;

[0014] The pixels corresponding to the next gray level to the right of the target gray level in the gray histogram are all taken as seed pixels.

[0015] Using seed pixels, the highlight area of ​​the local region is obtained according to the region growing algorithm; based on the shape of the highlight area, the first wrinkle analysis value of the local region is obtained.

[0016] The result of normalizing the negative number of the skewness coefficient of the gray-level histogram is used as the second wrinkle analysis value of the local region.

[0017] The normalized result of the negative correlation between the distribution concentration and the product of the first and second fold analysis values ​​is taken as the degree of folding in the local area.

[0018] Furthermore, the method for obtaining the degree of distribution concentration is as follows:

[0019] The difference between the target gray level and the minimum gray level in the gray level histogram is taken as the first difference, and the difference between the target gray level and the maximum gray level in the gray level histogram is taken as the second difference;

[0020] The difference between the first difference and the second difference is taken as the deviation analysis value; the product of the first difference and the deviation analysis value is negatively correlated and normalized, and the result is taken as the distribution concentration of the gray-level histogram.

[0021] Furthermore, the method for obtaining the first wrinkle analysis value is as follows:

[0022] For any highlighted area, obtain the skeleton of the highlighted area, and draw a tangent line through each pixel on the skeleton, which is used as a reference tangent line.

[0023] The length of the line segment formed by the intersection of each reference tangent and the highlighted area is used as the reference width;

[0024] The result of negatively correlating the product of the mean and standard deviation of the reference width is used as the width stability of the highlighted region.

[0025] The product of the proportion of pixels on the skeleton in the highlighted area and the width stability is used as the wrinkle reference degree of the highlighted area;

[0026] The sum of the wrinkle reference levels of all highlighted areas within the local region is used as the first wrinkle analysis value for that local region.

[0027] Furthermore, the method for obtaining the degree of reflectivity is as follows:

[0028] For any local area, obtain the non-highlighted area of ​​that local area as the reflectance analysis area;

[0029] Obtain the grayscale histograms corresponding to the grayscale values ​​of each row and column of pixels in the reflective analysis area, and use them as reference grayscale histograms;

[0030] In a two-dimensional coordinate system, gray level is used as the horizontal axis and frequency is used as the vertical axis. The curve corresponding to each reference gray level histogram in the two-dimensional coordinate system is obtained and used as the reference curve.

[0031] Arrange the reference curves corresponding to each row of pixels in the reflective analysis area in a top-to-bottom order to obtain a row curve sequence; obtain the row grayscale offset of each row of pixels based on the deviation between each reference curve in the row curve sequence and its previous adjacent reference curve.

[0032] Arrange the reference curves corresponding to each column of pixels in the reflective analysis area from left to right to obtain a column curve sequence; obtain the column grayscale offset of each column of pixels based on the deviation between each reference curve in the column curve sequence and its previous adjacent reference curve.

[0033] The sum of the standard deviations of all row grayscale offsets and all column grayscale offsets is negatively correlated and normalized to obtain the reflectivity of the local area.

[0034] Furthermore, the method for obtaining the row grayscale offset is as follows:

[0035] For the i-th reference curve and the (i-1)-th reference curve in the row curve sequence, the (i-1)-th reference curve is translated along the grayscale axis within a specified range. The difference between the (i-1)-th reference curve and the i-th reference curve after each translation is obtained through integration, and each difference is used as the first reference difference.

[0036] The translation amount of the (i-1)th reference curve corresponding to the smallest first reference difference is taken as the row grayscale offset of the row pixel corresponding to the i-th reference curve; where the row grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative.

[0037] Furthermore, the method for obtaining the column grayscale offset is as follows:

[0038] For the k-th reference curve and the (k-1)-th reference curve in the column curve sequence, the (k-1)-th reference curve is translated along the grayscale axis within a specified range. The difference between the (k-1)-th reference curve and the k-th reference curve after each translation is obtained through integration, and each difference is used as the second reference difference.

[0039] The translation amount of the (k-1)th reference curve corresponding to the smallest second reference difference is taken as the column grayscale offset of the column pixel corresponding to the kth reference curve; where the column grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative.

[0040] Furthermore, the method for correcting the grayscale value of pixels in each local region is as follows:

[0041] For any local region, arrange the row grayscale offsets in the local region according to the order of the corresponding rows from top to bottom to obtain the row grayscale offset sequence; add a 0 to the left end of the row grayscale offset sequence to construct the row grayscale offset vector;

[0042] Arrange the column grayscale offsets in the local area according to the order of the corresponding columns from left to right to obtain the column grayscale offset sequence; add a 0 to the left end of the column grayscale offset sequence to construct the column grayscale offset vector;

[0043] Multiply the row grayscale offset vector by the transpose of the column grayscale offset vector to obtain the grayscale correction matrix for the local area.

[0044] Map the grayscale value of each pixel in the local region to an empty matrix to obtain the grayscale matrix of the local region; where the number of rows in the grayscale matrix is ​​the number of rows of pixels in the local region, and the number of columns in the grayscale matrix is ​​the number of columns of pixels in the local region;

[0045] Each element in the difference matrix of the grayscale correction matrix is ​​subtracted from the grayscale matrix to obtain the corrected grayscale value of the corresponding pixel in the local region, thereby correcting the grayscale value of the pixel in the local region.

[0046] Furthermore, the method for obtaining the degree of wrinkle significance is as follows:

[0047] For any local area, the difference between the wrinkle degree obtained after gray value correction and the wrinkle degree before correction in the local area is normalized and used as the wrinkle degree of the local area.

[0048] The sum of the degree of wrinkling obtained after correcting the gray values ​​in the local area and the degree of wrinkle obviousness is used as the wrinkle significance analysis value of the local area.

[0049] The normalized product of the negative correlation result of reflectivity and the wrinkle significance analysis value is taken as the wrinkle significance of the local area.

[0050] Furthermore, the method for judging the stamping quality of metal foil parts based on the degree of wrinkle severity is as follows:

[0051] When the degree of wrinkle significance exceeds the preset wrinkle significance threshold, wrinkle defects exist in the corresponding local area;

[0052] When wrinkles are found in localized areas, the stamping quality of the metal foil parts is deemed unqualified.

[0053] When there are no wrinkles in any local area, the stamping quality of the metal foil parts is considered to be qualified.

[0054] The present invention has the following beneficial effects:

[0055] This invention first divides the image of the metal foil into local regions, which facilitates more accurate detection of defect areas. Then, based on the grayscale distribution of each local region, the degree of wrinkling in each region is obtained, initially determining the probability of wrinkle defects in each region. To avoid the influence of reflection and to more accurately identify wrinkle defects, the degree of reflection in each local region is obtained based on the grayscale distribution differences between adjacent rows and columns, accurately reflecting the reflection influence present in each region. Simultaneously, the grayscale values ​​of pixels in each local region are corrected to accurately remove the reflection influence, facilitating more accurate identification of wrinkle defects. Furthermore, based on the wrinkle degree and reflection degree before correction, the salience of wrinkles in each local region is obtained, accurately reflecting the probability of wrinkle defects in each region and effectively improving the accuracy of wrinkle defect identification. Finally, based on the salience of wrinkles, the quality of the stamping process of the metal foil parts is accurately judged, meeting the high-quality inspection requirements of metal foil parts in the precision manufacturing field. Attached Figure Description

[0056] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a schematic flowchart illustrating a quality inspection method for the stamping process of precision metal foil parts, provided as an embodiment of the present invention.

[0058] Figure 2 A flowchart illustrating a method for obtaining the degree of wrinkles according to an embodiment of the present invention;

[0059] Figure 3 This is a structural diagram of a quality inspection system for precision metal foil parts stamping, provided in one embodiment of the present invention.

[0060] Figure 4 This is a schematic diagram of a computer device provided according to an embodiment of the present invention. Detailed Implementation

[0061] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the quality inspection method for stamping precision metal foil parts according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0063] The following description, in conjunction with the accompanying drawings, details the specific scheme of the quality inspection method for the stamping process of precision metal foil parts provided by this invention.

[0064] Example 1:

[0065] The specific scenario of this embodiment is: for metal foils whose surfaces do not have defects that would damage the integrity of the foil, such as cracks or scratches.

[0066] This invention proposes a quality inspection method for the stamping process of precision metal foil parts. Please refer to [link / reference]. Figure 1 The diagram illustrates a schematic flowchart of a quality inspection method for stamping precision metal foil parts according to an embodiment of the present invention. The method includes the following steps:

[0067] Step S1: Obtain an image of the metal foil.

[0068] Specifically, the stamping process of the known metal foil is completed entirely on a horizontally arranged pallet. After the stamping process is finished, a dedicated quality inspection camera is directly activated to capture images of the processed metal foil without needing to move the foil, thus avoiding interference from secondary operations. It should be noted that the quality inspection camera is fixed directly above the pallet. This angle ensures that the shooting view covers the entire surface of the metal foil and minimizes image distortion caused by tilted shooting angles. This guarantees that the proportions of different areas of the metal foil in the image are consistent with reality, providing accurate spatial reference for subsequent defect location.

[0069] In practice, defect detection of metal foil relies heavily on surface grayscale differences (defects such as wrinkles and scratches alter local reflectivity, resulting in changes in grayscale values). To simplify subsequent analysis and focus on key features, the acquired metal foil images are converted to grayscale, ensuring that subsequent defect feature extraction revolves solely around grayscale data, thus improving detection efficiency and accuracy. It should be noted that all subsequent images are grayscale images of the metal foil. Grayscale conversion is a well-known technique and will not be elaborated upon further.

[0070] Step S2: Divide the image into local regions, and obtain the degree of wrinkling in each local region based on the distribution of gray values ​​in each local region.

[0071] Specifically, considering that defects in metal foil during the stamping process are few in reality, this embodiment first divides the image into various local regions to accurately and efficiently analyze these defects. Then, each local region is analyzed separately to improve the accuracy of defect identification. When the metal foil has no defects, the surface remains highly flat, and the color within the local region is uniform with no obvious visual differences. When the metal foil has cracks and scratches, these defects directly damage its integrity. Due to the natural color difference between the metal foil and the tray, a clear visual boundary is formed, allowing cracks and scratches to be quickly and accurately identified. Unlike cracks and scratches, wrinkles are essentially non-uniform plastic deformation of the metal foil, manifested as morphological changes such as protrusions and depressions on the surface. However, the material and color of the metal foil itself remain unchanged, always consistent with the normal area. Therefore, the detection of wrinkles relies on differences in light reflection. Because protrusions and depressions change the local light reflection path, the amount of light entering the camera fluctuates, ultimately manifesting as differences in grayscale values ​​in the image.

[0072] However, due to the ultra-thin nature of metal foil, the bulges and depressions caused by wrinkles are very slight, and the corresponding grayscale value differences are extremely weak, making it difficult to distinguish from the grayscale changes in normal areas, resulting in low detection accuracy of wrinkle defects. At the same time, the ultra-thin nature of metal foil also makes it extremely sensitive to light, and the surface of metal foil is more prone to obvious reflection. The reflection will form local bright areas in the image, and the grayscale value changes of these areas are highly similar to the grayscale differences caused by wrinkle defects, thus affecting the overall accuracy of metal foil material inspection. Considering that the morphological changes of wrinkles in metal foil will exhibit unique distribution patterns in grayscale images due to differences in light reflection, specifically, the surface normal direction of the top of the protrusion is consistent with that of the normal foil area, allowing light to be reflected normally into the camera. Furthermore, the surface of the top of the protrusion is smoother, with a slightly higher reflectivity than the normal area, resulting in more light entering the lens and a higher grayscale value. The surface normal directions on both sides of the protrusion deflect to the left and right respectively, allowing only a small amount of diffused light to enter the camera, significantly reducing the amount of light entering and causing the corresponding grayscale value to be lower than the normal area. When incident light enters the recessed area, it undergoes multiple reflections on the inner wall of the recess, preventing most of the light from escaping. Even if a small amount of light does escape, its energy is significantly attenuated, resulting in very little light ultimately entering the camera, thus causing the grayscale value of the recessed area to be significantly lower than the normal area. Therefore, wrinkles appear as narrow lines in an image. When a wrinkle exists in a localized area, the gray-level histogram distribution within that area exhibits a unique unimodal skewness, deviating further from a normal distribution. The peak in the gray-level histogram corresponds to the high-gray-level area at the top of the bulge within the local area. Considering the small overall proportion of the wrinkle, the number of high-gray-level pixels should be extremely small; therefore, the number of pixels decreases rapidly as the peak extends towards increasing gray-level. Since the sides of the bulge and the recessed areas exhibit low-gray-level characteristics, the number of low-gray-level pixels is relatively large; therefore, the number of pixels decreases slowly as the peak extends towards decreasing gray-level. Overall, the gray-level histogram of the wrinkled area exhibits a unimodal skewness distribution, meaning that based on the gray-level values ​​of the normal area, the peak is biased towards the high-gray-level end (corresponding to the top of the bulge), while the low-gray-level end shows a trailing characteristic (corresponding to the sides of the bulge and the recessed areas). The method for obtaining the gray-level histogram is well-known and will not be elaborated further.

[0073] Therefore, in order to analyze the wrinkling situation in each local area, this embodiment obtains the wrinkling degree of each local area based on the distribution of grayscale values ​​in each local area. The greater the wrinkling degree, the more likely the corresponding local area is to have wrinkling defects. It should be noted that the size of the local area is set in this embodiment as [missing information]. The implementer can set the size of the local region according to the actual situation, and there is no limit here. If the boundary region of the image is not enough to form a complete local region, the remaining boundary region is considered as a local region by default.

[0074] Preferably, in one feasible embodiment, the method for obtaining the degree of wrinkles is described in [reference needed]. Figure 2 The document presents a flowchart of a method for obtaining the degree of wrinkles provided in this embodiment, which includes the following steps:

[0075] Step S201: Obtain the degree of distribution concentration.

[0076] For any given local region, its grayscale histogram is obtained. The gray level corresponding to the largest quantity in the local region's grayscale histogram is taken as the target gray level. Then, the position of the target gray level in the local region's grayscale histogram is analyzed. The closer the target gray level is to the middle of the grayscale range corresponding to the gray level in the local region's grayscale histogram, the more normally the grayscale histogram distribution of the local region conforms to, indirectly indicating that the local region is less likely to have wrinkle defects. Furthermore, this embodiment obtains the distribution concentration of the grayscale histogram based on the position distribution of the target gray level in the grayscale histogram. The greater the distribution concentration, the less likely the local region is to have wrinkle defects. For better explanation, the following analysis will use this local region as an example.

[0077] The method for obtaining the distribution concentration is as follows: the absolute value of the difference between the target gray level and the smallest gray level in the gray-level histogram is taken as the first difference, and the absolute value of the difference between the target gray level and the largest gray level in the gray-level histogram is taken as the second difference. When the first difference is larger than the second difference, the distribution of the gray-level histogram is less likely to conform to a normal distribution, indirectly indicating that the distribution of the gray-level histogram is more deviated. Therefore, the difference between the first difference and the second difference is taken as the deviation analysis value. Then, the product of the first difference and the deviation analysis value is negatively correlated and normalized, and the result is taken as the distribution concentration of the gray-level histogram. In this embodiment, the result of normalizing the product of the first difference and the deviation analysis value is taken as the first result, and then the difference between the constant 1 and the first result is taken as the result of negatively correlated and normalized the product of the first difference and the deviation analysis value. The normalization process for the product of the first difference and the deviation analysis value is performed using the norm normalization function.

[0078] Step S202: Obtain the first wrinkle analysis value.

[0079] It is known that wrinkles appear as narrow lines in an image and are always highlighted areas. Therefore, in this embodiment, the pixels corresponding to the next gray level to the right of the target gray level in the gray-scale histogram of this local area are all used as seed pixels for this local area. Then, using the seed pixels, a region growing algorithm is used to obtain the generated region. The growth condition of the region growing algorithm is: growth proceeds in the direction of increasing gray value, that is, the pixel with the largest gray value in the 8-neighborhood of the seed pixel is selected and added to the growing region. If there are multiple pixels with the largest gray values ​​in the 8-neighborhood of the seed pixel, then all of them are added to the growing region. The region growing algorithm is a well-known technique and will not be described in detail here. To eliminate noise interference, this embodiment further obtains the number of pixels in each generated region as the first quantity for each generated region. When the first quantity is greater than a preset first quantity threshold, the corresponding generated region is designated as the highlight region of that local region. In this embodiment, the preset first quantity threshold is set to 4. Implementers can set the size of the preset first quantity threshold according to actual conditions, which is not limited here. The more the shape of the highlight region conforms to a narrow line, the more likely the local region is to have wrinkle defects. Therefore, this embodiment obtains the first wrinkle analysis value of the local region based on the shape of the highlight region. The larger the first wrinkle analysis value, the more likely the local region is to have wrinkle defects.

[0080] In one possible implementation of this embodiment, the method for obtaining the first wrinkle analysis value is as follows: for any bright area, the skeleton of the bright area is obtained by a skeleton extraction algorithm, and a tangent line is drawn through each pixel on the skeleton, which is used as a reference tangent line; wherein, the skeleton extraction algorithm is a well-known technology and will not be described in detail. Then, the length of the line segment formed by the intersection of each reference tangent and the highlighted area is used as the reference width. The smaller and more consistent the reference widths are, the more the highlighted area conforms to a narrow line shape, and the more likely it is to be a wrinkled area. At the same time, the larger the proportion of pixels on the skeleton in the highlighted area, the more the highlighted area conforms to a narrow line shape, and the more likely it is to be a wrinkled area. Furthermore, in this embodiment, the product of the mean and standard deviation of the reference width is negatively correlated to determine the width stability of the highlighted area. The greater the width stability, the smaller and more stable the reference width, and the more likely the highlighted area is to be a wrinkled area. In this embodiment, the negative of the product of the mean and standard deviation of the reference width is used as the power of an exponential function with the natural constant as the base. The output of this exponential function is the result of negatively correlated product of the mean and standard deviation of the reference width. Then, the product of the proportion of pixels on the skeleton in the highlighted area and the width stability is used as the wrinkle reference degree of the highlighted area; in order to analyze the wrinkle situation of the local area as a whole, this embodiment adds up the wrinkle reference degree of all highlighted areas in the local area and uses it as the first wrinkle analysis value of the local area.

[0081] Step S203: Obtain the second wrinkle analysis value.

[0082] It is known that when the gray-level histogram of a local region is left-skewed, the local region is more likely to have wrinkle defects. Therefore, this embodiment obtains the skewness coefficient of the gray-level histogram of the local region. The smaller the negative skewness coefficient, the more likely the local region is to have wrinkle defects. This embodiment then normalizes the negative value of the skewness coefficient of the gray-level histogram of the local region as the second wrinkle analysis value for that local region. The larger the second wrinkle analysis value, the more likely the local region is to have wrinkle defects. This embodiment normalizes the negative value of the skewness coefficient using a norm normalization function. The method for obtaining the skewness coefficient is a well-known technique and will not be described in detail here.

[0083] Step S204: Obtain the degree of wrinkles.

[0084] It is known that the greater the concentration of distribution, the less likely the local area is to have wrinkle defects; the larger the first wrinkle analysis value, the more likely the local area is to have wrinkle defects; the larger the second wrinkle analysis value, the more likely the local area is to have wrinkle defects. Therefore, in this embodiment, the normalized result of the negative correlation between the distribution concentration and the product of the first and second wrinkle analysis values ​​is taken as the wrinkle degree of the local area. In this embodiment, the reciprocal of the sum of the distribution concentration and a first preset constant is taken as the negative correlation result of the distribution concentration; the negative correlation result of the distribution concentration and the product of the first and second wrinkle analysis values ​​are normalized using a norm normalization function. The first preset constant is a positive number to avoid a denominator of 0; in this embodiment, the first preset constant is set to 1. Implementers can set the size of the first preset constant according to actual conditions, and this is not limited here.

[0085] At this point, the degree of wrinkling in each local area is obtained.

[0086] Step S3: Based on the grayscale distribution differences between adjacent rows and columns in each local area, obtain the reflectivity of each local area and correct the grayscale value of the pixels in each local area.

[0087] Specifically, in the quality inspection of stamping processes for metal foil parts, reflection and wrinkle defects may coexist, meaning they overlap in image grayscale distribution, thus affecting the accuracy of wrinkle defect identification. To more accurately identify wrinkle defects in metal foil, the essential differences in grayscale change characteristics between reflection and wrinkle defects can be utilized. Specifically: Reflection causes a clear directionality in the grayscale changes of metal foil images; that is, under illumination, the grayscale value in a certain direction of the image will show a regular gradual change. For example, in a certain local area, from left to right, the grayscale value will continuously increase due to the gradually increasing intensity of reflected light. For instance, when the reflection is more severe on the right side of the local area, the grayscale value in that local area will gradually increase from left to right. The grayscale change of wrinkle defects mainly stems from the difference in the amount of reflected light caused by local plastic deformation (protrusions and depressions) of the foil. Its grayscale distribution is more concentrated in the deformation area and does not have the regular characteristic of gradual change along a specific direction. Therefore, by capturing the directional features of image grayscale changes, a crucial basis can be provided for subsequent elimination of reflective interference and optimization of wrinkle defect judgment. Furthermore, this embodiment obtains the reflectivity of each local region based on the grayscale distribution differences between adjacent rows and columns in each local region and corrects the grayscale values ​​of pixels in each local region. The greater the reflectivity, the greater the impact of reflective interference on the corresponding local region.

[0088] Preferably, in one feasible embodiment, the method for obtaining the reflectivity is as follows: for any local area, the non-high-brightness area of ​​that local area is obtained as the reflectivity analysis area. The non-high-brightness area is chosen as the reflectivity analysis area because the cause of high-brightness areas is uncertain; that is, high-brightness areas may be wrinkles caused by local plastic deformation of the metal foil, or they may be specular reflection interference generated by the ultra-smooth foil surface. Before clearly distinguishing between wrinkle high-brightness and reflective high-brightness, it is impossible to accurately determine whether the high-brightness area is a real wrinkle defect. If the high-brightness area is also used as the reflectivity analysis area, it will cause interference information to be mixed into the grayscale feature analysis of the normal foil, thereby affecting the accuracy of subsequent reflectivity judgment, image correction, and wrinkle defect probability calculation. Therefore, selecting the non-high-brightness area as the reflectivity analysis area can ensure to the greatest extent that the analysis object is an undisputed potential normal area, providing a reliable data foundation for subsequent grayscale distribution feature extraction, and facilitating a more accurate analysis of the reflectivity influence existing in that local area.

[0089] To analyze the reflection in this local area, gray-level histograms corresponding to the gray values ​​of each row and column of pixels in the reflection analysis area are obtained and used as reference gray-level histograms. Then, in a two-dimensional coordinate system, gray level is used as the horizontal axis and frequency as the vertical axis to obtain the curve corresponding to each reference gray-level histogram in the two-dimensional coordinate system, which is used as a reference curve. Then, the offset between the reference curves between rows and columns is analyzed to more accurately analyze the reflection effect on this local area.

[0090] First, the reference curves corresponding to each row of pixels in the reflective analysis area are arranged in a top-to-bottom order to obtain a row curve sequence. Then, based on the deviation between each reference curve in the row curve sequence and its preceding adjacent reference curve, the row grayscale offset of each row of pixels is obtained. Specifically, the row grayscale offset is obtained as follows: for the i-th reference curve and the (i-1)-th reference curve in the row curve sequence, the (i-1)-th reference curve is translated along the grayscale axis within a specified range. The absolute value of the difference between the (i-1)-th reference curve and the i-th reference curve after each translation is obtained through integration, and this difference is used as the first reference difference. In this embodiment, the specified range is defined as... The implementer can set the specified range of movement according to the actual situation, which is not limited here. The step size of the (i-1)th reference curve in each translation is 1. The translation amount of the (i-1)th reference curve corresponding to the smallest first reference difference is taken as the row grayscale offset of the pixel corresponding to the i-th reference curve; the row grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative. Thus, the row grayscale offset of the pixel corresponding to each reference curve in the row curve sequence is obtained. It should be noted that because the first reference curve in the row curve sequence does not have a preceding adjacent reference curve, the row grayscale offset of the pixel corresponding to the first reference curve is not obtained.

[0091] Further, the reference curves corresponding to each column of pixels in the reflective analysis area are arranged from left to right to obtain a column curve sequence. Based on the deviation between each reference curve in the column curve sequence and its preceding adjacent reference curve, the column grayscale offset of each column of pixels is obtained. The method for obtaining the column grayscale offset is as follows: for the k-th reference curve and the (k-1)-th reference curve in the column curve sequence, the (k-1)-th reference curve is translated along the grayscale axis within a specified range. The absolute value of the difference between the (k-1)-th reference curve and the k-th reference curve after each translation is obtained through integration, and this value is used as the second reference difference. In this embodiment, the specified range is set to be... The implementer can set the specified range of movement according to the actual situation, which is not limited here. The step size of the (k-1)th reference curve in each translation is 1. The translation amount of the (k-1)th reference curve corresponding to the smallest second reference difference is taken as the column grayscale offset of the corresponding column pixel of the kth reference curve; the column grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative. Thus, the column grayscale offset of the corresponding column pixel of each reference curve in the column curve sequence is obtained. It should be noted that because the first reference curve in the column curve sequence does not have a preceding adjacent reference curve, the column grayscale offset of the corresponding column pixel of the first reference curve is not obtained.

[0092] When the fluctuations in both row and column grayscale offsets are more stable, it indicates a greater degree of reflection in the local area. Therefore, in this embodiment, the sum of the standard deviations of all row and column grayscale offsets is negatively correlated and normalized to represent the reflectivity of the local area. This embodiment uses the negative of the sum of the standard deviations as the power of an exponential function with a base of the natural constant. The output of this exponential function is the sum of the standard deviations negatively correlated and normalized. It should be noted that if all pixels in a row or column of a local area are in the highlighted area, it means that there are at least two reflection analysis areas in that local area. In this case, each reflection analysis area needs to be analyzed independently first to obtain the row grayscale offset of each row and the column grayscale offset of each column in each reflection analysis area, and then merged. It should also be noted that, except for the first row and the first column of the local area, the row grayscale offsets of other rows for which the row grayscale offsets cannot be obtained are all set to 0, and the column grayscale offsets of other columns for which the column grayscale offsets cannot be obtained are also all set to 0.

[0093] At this point, the reflectivity of each local area is obtained.

[0094] Preferably, in one feasible method of this embodiment, the method for correcting the grayscale value of pixels in each local region is as follows: For any local region, the row grayscale offsets in the local region are arranged according to the order of the corresponding rows from top to bottom to obtain a row grayscale offset sequence; then, a 0 is added to the left end of the row grayscale offset sequence to construct a row grayscale offset vector. It should be noted that the reason for adding 0 is to consider that the first row of pixels in the local region does not have a corresponding row grayscale offset. Further, the column grayscale offsets in the local region are arranged according to the order of the corresponding columns from left to right to obtain a column grayscale offset sequence; then, a 0 is added to the left end of the column grayscale offset sequence to construct a column grayscale offset vector. It should be noted that the reason for adding 0 is to consider that the first column of pixels in the local region does not have a corresponding column grayscale offset.

[0095] Multiply the row grayscale offset vector by the transpose of the column grayscale offset vector to obtain the grayscale correction matrix for the local region. Simultaneously, map the grayscale value of each pixel in the local region to an empty matrix to obtain the grayscale matrix for the local region. The number of rows in the grayscale matrix is ​​the same as the number of rows and columns in the local region. Finally, subtract each element of the difference matrix of the grayscale correction matrix from the grayscale matrix to obtain the corrected grayscale value of the corresponding pixel in the local region. This corrects the grayscale value of the pixels in the local region, thus removing the reflective effect in the local region.

[0096] At this point, the grayscale values ​​of pixels in each local region are corrected.

[0097] Step S4: Based on the degree of wrinkles and the degree of reflection of each local area before correction, obtain the degree of wrinkle significance of each local area.

[0098] Specifically, the greater the degree of wrinkling obtained after grayscale value correction in a local area, and the more significantly it is compared to the degree of wrinkling before correction, the more accurately it indicates the presence of wrinkle defects in that local area. Therefore, this embodiment first analyzes the degree of wrinkle defects in each local area based on its degree of wrinkling before correction. Considering that a higher degree of reflectivity in a local area indicates a greater influence from reflection, the analysis results of the degree of wrinkle defects in that local area can be further corrected based on the degree of reflectivity, thus making the analysis of whether wrinkle defects exist in that local area more accurate. Therefore, this embodiment obtains the degree of wrinkle significance in each local area based on its degree of wrinkling before correction and its degree of reflectivity. The greater the degree of wrinkle significance, the more likely wrinkle defects exist in the corresponding local area.

[0099] Preferably, in one feasible embodiment of this invention, the method for obtaining the salience of wrinkles is as follows: for any local area, the difference between the wrinkle degree obtained after grayscale value correction and the wrinkle degree before correction in the local area is normalized, and this normalized result is taken as the wrinkle salience of the local area; the greater the wrinkle salience, the more accurate the wrinkle analysis of the local area. Considering that when there is less reflective influence in the local area, if there is a wrinkle defect in the local area, the wrinkle salience of the local area will be smaller, in order to more accurately characterize the possibility of wrinkle defects in the local area, this embodiment further adds the wrinkle degree obtained after grayscale value correction and the wrinkle salience in the local area as the wrinkle salience analysis value of the local area; the greater the wrinkle salience analysis value, the more likely there is a wrinkle defect in the local area;

[0100] Considering that the degree of reflectivity can indirectly reflect the accuracy of wrinkle defect analysis in a local area—that is, the lower the degree of reflectivity, the more accurate the wrinkle defect analysis, and the higher the degree of reflectivity, the less accurate the wrinkle defect analysis—the product of the negative correlation result of reflectivity and the wrinkle significance analysis value is normalized and used as the wrinkle significance degree of the local area. In this embodiment, the negative number of reflectivity is used as the power of an exponential function with the natural constant as the base. The output of this exponential function is the negative correlation result of reflectivity. The product of the negative correlation result of reflectivity and the wrinkle significance analysis value is normalized using a norm normalization function.

[0101] At this point, the salience of wrinkles in each local area is obtained.

[0102] Step S5: Determine the quality of stamping of metal foil parts based on the degree of wrinkle severity.

[0103] It is known that the greater the degree of wrinkle significance, the more likely there is a wrinkle defect in the corresponding local area, which indirectly indicates that there is a wrinkle defect in the metal foil, and further indicates that the stamping quality of the metal foil parts is unqualified. Therefore, this embodiment judges the stamping quality of metal foil parts based on the degree of wrinkle significance.

[0104] The method for judging the quality of stamping of metal foil parts based on the degree of wrinkle severity is as follows: First, a preset wrinkle severity threshold of 0.5 is set. The implementer can set the size of the preset wrinkle severity threshold according to the actual situation, which is not limited here. When the wrinkle severity is greater than the preset wrinkle severity threshold, a wrinkle defect exists in the corresponding local area, that is, the highlighted area of ​​the corresponding local area is the wrinkle defect area. Therefore, when a wrinkle defect is found in a local area, it indicates that the metal foil has a wrinkle defect, and the stamping quality of the metal foil parts is judged to be unqualified. When no wrinkle areas exist in any local area, it indicates that the metal foil does not have a wrinkle defect, and the stamping quality of the metal foil parts is judged to be qualified.

[0105] In summary, this embodiment acquires an image of a metal foil; based on the distribution of grayscale values ​​in each local region of the image, it obtains the degree of wrinkling in each local region; based on the difference in grayscale distribution between adjacent rows and columns in each local region, it obtains the degree of reflectivity in each local region and corrects the grayscale values ​​of pixels in each local region; based on the degree of wrinkling and reflectivity of each local region before correction, it obtains the salience of wrinkles in each local region and judges the quality of the stamping process of the metal foil parts. This invention, by obtaining the salience of wrinkles, accurately determines whether wrinkles exist in corresponding local regions, effectively improving the accuracy of wrinkle defect identification in metal foil and facilitating a more accurate judgment of the quality of the stamping process of metal foil parts.

[0106] Example 2:

[0107] This invention also proposes a quality inspection system for the stamping process of precision metal foil parts; please refer to [link to relevant documentation]. Figure 3 The diagram shows a structural diagram of a quality inspection system for precision metal foil parts stamping provided by an embodiment of the present invention. The system includes: an image acquisition module 10, a wrinkle degree acquisition module 20, a reflectivity analysis module 30, a wrinkle significance degree acquisition module 40, and a quality inspection module 50.

[0108] Image acquisition module 10 is used to acquire images of metal foil.

[0109] The wrinkle degree acquisition module 20 is used to divide the image into local regions and acquire the wrinkle degree of each local region based on the distribution of gray values ​​in each local region.

[0110] The reflectivity analysis module 30 is used to obtain the reflectivity of each local area based on the grayscale distribution differences between adjacent rows and adjacent columns in each local area and to correct the grayscale values ​​of pixels in each local area.

[0111] The wrinkle salience acquisition module 40 is used to acquire the wrinkle salience of each local area based on the wrinkle degree before correction and the reflectivity of each local area.

[0112] The quality inspection module 50 is used to judge the quality of stamping of metal foil parts based on the degree of wrinkle severity.

[0113] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the quality inspection system for stamping precision metal foil parts and the quality inspection method for stamping precision metal foil parts provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiments and will not be repeated here.

[0114] Example 3:

[0115] This invention also proposes a quality inspection device for the stamping process of precision metal foil parts. The device includes a memory and a processor. The memory stores executable program code, and the processor calls and executes the executable program code to perform a quality inspection method for the stamping process of precision metal foil parts provided in the embodiments of this application. Specifically, the device may be a chip, component, or module. The chip may include a connected processor and memory; the memory stores instructions, and when the processor calls and executes the instructions, the chip can perform the quality inspection method for the stamping process of precision metal foil parts provided in the above embodiments.

[0116] Furthermore, this application also protects a computer device; please refer to [link to relevant documentation]. Figure 4 The computer device includes a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402. When the processor 402 executes the computer program 403, the computer device can perform any of the aforementioned quality inspection methods for stamping precision metal foil parts.

[0117] Example 4:

[0118] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the quality inspection method for stamping precision metal foil parts provided in the above embodiment.

[0119] Example 5:

[0120] This embodiment also provides a computer program product. When the computer program product is run on a computer, it causes the computer to perform the above-mentioned related steps to realize the quality inspection method for stamping of precision metal foil parts provided in the above embodiment.

[0121] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0122] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0123] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A quality inspection method for stamping precision metal foil parts, characterized in that, The method includes the following steps: Acquire images of metal foil; The image is divided into local regions, and the degree of wrinkling in each local region is obtained based on the distribution of gray values ​​in each local region. Based on the difference in grayscale distribution between adjacent rows and columns in each local region, the reflectivity of each local region is obtained and the grayscale value of the pixels in each local region is corrected; Based on the degree of wrinkling and reflectivity of each local area before correction, the degree of wrinkling significance of each local area is obtained; The quality of stamping of metal foil parts is judged based on the degree of wrinkle. The method for obtaining the degree of reflectivity is as follows: For any local area, obtain the non-highlighted area of ​​that local area as the reflectance analysis area; Obtain the grayscale histograms corresponding to the grayscale values ​​of each row and column of pixels in the reflective analysis area, and use them as reference grayscale histograms; In a two-dimensional coordinate system, gray level is used as the horizontal axis and frequency is used as the vertical axis. The curve corresponding to each reference gray level histogram in the two-dimensional coordinate system is obtained and used as the reference curve. Arrange the reference curves corresponding to each row of pixels in the reflective analysis area in a top-to-bottom order to obtain a row curve sequence; obtain the row grayscale offset of each row of pixels based on the deviation between each reference curve in the row curve sequence and its previous adjacent reference curve. Arrange the reference curves corresponding to each column of pixels in the reflective analysis area from left to right to obtain a column curve sequence; obtain the column grayscale offset of each column of pixels based on the deviation between each reference curve in the column curve sequence and its previous adjacent reference curve. The sum of the standard deviations of all row grayscale offsets and all column grayscale offsets is negatively correlated and normalized to obtain the reflectivity of the local area. The method for obtaining the degree of wrinkle significance is as follows: For any local area, the difference between the wrinkle degree obtained after gray value correction and the wrinkle degree before correction in the local area is normalized and used as the wrinkle degree of the local area. The sum of the degree of wrinkling obtained after correcting the gray values ​​in the local area and the degree of wrinkle obviousness is used as the wrinkle significance analysis value of the local area. The normalized product of the negative correlation result of reflectivity and the wrinkle significance analysis value is taken as the wrinkle significance of the local area.

2. The quality inspection method for stamping precision metal foil parts as described in claim 1, characterized in that, The method for obtaining the degree of wrinkles is as follows: For any local region, the gray level corresponding to the largest number in the gray-level histogram of that local region is taken as the target gray level; Based on the positional distribution of the target gray level in the gray-level histogram, the degree of concentration of the gray-level histogram is obtained; The pixels corresponding to the next gray level to the right of the target gray level in the gray histogram are all taken as seed pixels. Using seed pixels, the highlight area of ​​the local region is obtained according to the region growing algorithm; based on the shape of the highlight area, the first wrinkle analysis value of the local region is obtained. The result of normalizing the negative number of the skewness coefficient of the gray-level histogram is used as the second wrinkle analysis value of the local region. The normalized result of the negative correlation between the distribution concentration and the product of the first and second fold analysis values ​​is taken as the degree of folding in the local area.

3. The quality inspection method for stamping precision metal foil parts as described in claim 2, characterized in that, The method for obtaining the degree of distribution concentration is as follows: The difference between the target gray level and the minimum gray level in the gray level histogram is taken as the first difference, and the difference between the target gray level and the maximum gray level in the gray level histogram is taken as the second difference; The difference between the first difference and the second difference is taken as the deviation analysis value; the product of the first difference and the deviation analysis value is negatively correlated and normalized, and the result is taken as the distribution concentration of the gray-level histogram.

4. A quality inspection method for stamping precision metal foil parts as described in claim 2, characterized in that, The method for obtaining the first wrinkle analysis value is as follows: For any highlighted area, obtain the skeleton of the highlighted area, and draw a tangent line through each pixel on the skeleton, which is used as a reference tangent line. The length of the line segment formed by the intersection of each reference tangent and the highlighted area is used as the reference width; The result of negatively correlating the product of the mean and standard deviation of the reference width is used as the width stability of the highlighted region. The product of the proportion of pixels on the skeleton in the highlighted area and the width stability is used as the wrinkle reference degree of the highlighted area; The sum of the wrinkle reference levels of all highlighted areas within the local region is used as the first wrinkle analysis value for that local region.

5. A quality inspection method for stamping precision metal foil parts as described in claim 1, characterized in that, The method for obtaining the row grayscale offset is as follows: For the i-th reference curve and the (i-1)-th reference curve in the row curve sequence, the (i-1)-th reference curve is translated along the grayscale axis within a specified range. The difference between the (i-1)-th reference curve and the i-th reference curve after each translation is obtained through integration, and each difference is used as the first reference difference. The translation amount of the (i-1)th reference curve corresponding to the smallest first reference difference is taken as the row grayscale offset of the row pixel corresponding to the i-th reference curve; where the row grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative.

6. A quality inspection method for stamping precision metal foil parts as described in claim 1, characterized in that, The method for obtaining the column grayscale offset is as follows: For the k-th reference curve and the (k-1)-th reference curve in the column curve sequence, the (k-1)-th reference curve is translated along the grayscale axis within a specified range. The difference between the (k-1)-th reference curve and the k-th reference curve after each translation is obtained through integration, and each difference is used as the second reference difference. The translation amount of the (k-1)th reference curve corresponding to the smallest second reference difference is taken as the column grayscale offset of the column pixel corresponding to the kth reference curve; where the column grayscale offset can be positive or negative, with rightward shift being positive and leftward shift being negative.

7. A quality inspection method for stamping precision metal foil parts as described in claim 1, characterized in that, The method for correcting the grayscale value of pixels in each local region is as follows: For any local region, arrange the row grayscale offsets in the local region according to the order of the corresponding rows from top to bottom to obtain the row grayscale offset sequence; add a 0 to the left end of the row grayscale offset sequence to construct the row grayscale offset vector; Arrange the column grayscale offsets in the local area according to the order of the corresponding columns from left to right to obtain the column grayscale offset sequence; add a 0 to the left end of the column grayscale offset sequence to construct the column grayscale offset vector; Multiply the row grayscale offset vector by the transpose of the column grayscale offset vector to obtain the grayscale correction matrix for the local area. Map the grayscale value of each pixel in the local region to an empty matrix to obtain the grayscale matrix of the local region; where the number of rows in the grayscale matrix is ​​the number of rows of pixels in the local region, and the number of columns in the grayscale matrix is ​​the number of columns of pixels in the local region; Each element in the difference matrix of the grayscale correction matrix is ​​subtracted from the grayscale matrix to obtain the corrected grayscale value of the corresponding pixel in the local region, thereby correcting the grayscale value of the pixel in the local region.

8. A quality inspection method for stamping precision metal foil parts as described in claim 1, characterized in that, The method for judging the quality of stamping of metal foil parts based on the degree of wrinkle severity is as follows: When the degree of wrinkle significance exceeds the preset wrinkle significance threshold, wrinkle defects exist in the corresponding local area; When wrinkles are found in localized areas, the stamping quality of the metal foil parts is deemed unqualified. When there are no wrinkles in any local area, the stamping quality of the metal foil parts is considered to be qualified.

Images