Oil stone static pressure forming quality detection method based on image processing

Abstract

The invention relates to the field of image processing, in particular to an oilstone static pressure forming quality detection method based on image processing, and the method comprises the steps: obtaining an infrared image set of an oilstone sintering and cooling stage, carrying out the superpixel segmentation to obtain a superpixel region of each frame of image, matching the superpixel regions of adjacent frames, and building a corresponding relation, and calculating a temperature anomaly characteristic value of each region. And analyzing the temperature change trend of each region, calculating a correlation coefficient, and introducing distance and temperature anomaly weight to carry out weighted fusion to obtain a comprehensive correlation index. And in combination with the temperature anomaly characteristic value and the comprehensive correlation index, evaluating the possibility value of occurrence of pore anomaly in each superpixel region, and judging whether pore distribution anomaly exists in the oilstone or not, thereby completing quality detection. According to the method, through superpixel segmentation and matching, in combination with temperature anomaly characteristics and comprehensive correlation analysis, precise detection of oilstone internal pore anomaly is realized, and the reliability of static pressure forming quality detection is improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to an image processing-based method for detecting the quality of oilstone static pressure forming. Background Art

[0002] Whetstone is a grinding tool made of abrasives (such as aluminum oxide, silicon carbide, etc.) and a binder. It is used to finely polish, deburr or trim the surface of a workpiece through manual or mechanical friction. It is used in mechanical manufacturing, bearing processing, tool sharpening and other fields. It is an indispensable tool in precision machining.

[0003] In the production process of oilstone, the molding process is very important, and static pressure molding is an ideal choice for oilstones with high density uniformity requirements because it can avoid the density gradient problem of traditional molding. However, the internal pores of the oilstone formed by static pressure molding may be crushed due to excessive static pressure, or the pores may be too large or unevenly distributed due to insufficient pressure. These problems will affect the heat dissipation, chip removal or lubrication performance of the oilstone, and thus affect its processing effect and service life. Therefore, it is very important to test the quality of static pressure molding of oilstones.

[0004] Traditional detection methods have many limitations when detecting abnormal internal pore distribution after oilstone static pressing. Nondestructive testing methods such as ultrasonic testing and X-ray testing have low resolution, making it difficult to accurately identify small pores, and are easily disturbed by the internal structural noise of the oilstone, resulting in insufficient detection accuracy and slow detection speed. They are difficult to adapt to large-scale production, and cannot monitor quality changes during the production process in real time, making it difficult to adjust process parameters in a timely manner. In addition, for irregularly shaped oilstones, traditional methods are difficult to fully cover the internal area, and are highly dependent on the experience of the inspectors, which can easily lead to misjudgments or missed judgments. Summary of the invention

[0005] In order to solve the problems of low resolution, insufficient accuracy, slow speed, inability to monitor in real time, poor adaptability to irregular shapes, and easy misjudgment and missed judgment due to reliance on experience when detecting abnormal internal pore distribution of oilstone after static pressing, the present invention provides solutions in the following aspects.

[0006] A method for detecting the quality of oilstone static pressing molding based on image processing comprises: obtaining an infrared image set of the oilstone sintering cooling stage and performing preprocessing; performing superpixel segmentation according to each frame of the oilstone image in the infrared image set to obtain superpixel regions, performing superpixel region matching on two adjacent frames of the oilstone image in the infrared image set in turn to establish a corresponding relationship, and calculating the temperature anomaly characteristic value of each superpixel region in each frame of the oilstone image during the cooling process according to the matching result; analyzing the temperature change trend of each superpixel region in different frames of the oilstone image, calculating the correlation coefficient between each superpixel region and other superpixel regions, and introducing distance weights and temperature anomaly weights for weighted fusion to obtain a comprehensive correlation index, and then combining the temperature anomaly characteristic value to evaluate the possibility value of whether pore anomalies occur inside each superpixel region; judging whether there is pore distribution anomaly inside the oilstone according to the possibility value of pores appearing in each superpixel region, and completing the quality detection target of oilstone static pressing molding.

[0007] The effect is as follows: by performing superpixel segmentation and matching on the infrared image set, combining the temperature anomaly eigenvalues ​​with the comprehensive correlation index, it is possible to accurately evaluate the possibility of pore anomalies in each superpixel area, effectively discover potential defects inside the oilstone, and use superpixel segmentation and matching technology to quickly establish the correspondence between superpixel areas in adjacent frame images, improve detection efficiency, and ensure timely discovery of quality problems. Taking into account the temperature anomaly eigenvalues ​​and correlation indicators, the pore distribution inside the oilstone is comprehensively evaluated to provide a more accurate basis for quality detection. By analyzing the temperature change trend and correlation, and combining the distance weight and the temperature anomaly weight for weighted fusion, the detection results are more convincing and reliable, and the abnormal pore distribution inside the oilstone is discovered in time, which helps to adjust the process parameters in time during the production process, improve the performance and service life of the oilstone, and ensure stable and reliable product quality.

[0008] Preferably, the super pixel area includes: Obtain the grayscale values ​​of all pixels in each frame of the oilstone image, and use the Sobel operator to calculate the gradient amplitude of each pixel; Based on the grayscale value, gradient value and position coordinates of the pixel points in each frame of the oilstone image, the SLIC superpixel segmentation algorithm is used to perform superpixel segmentation on the infrared image set to obtain the superpixel area.

[0009] Preferably, the step of sequentially performing superpixel region matching on two adjacent oilstone images in the infrared image set includes: Taking any oilstone image as the target frame image, obtaining the coordinates of the central pixel point of the superpixel region in the target frame image, and calculating the Euclidean distance between the central pixel point of the superpixel region in the target frame image and the next oilstone image to match the corresponding area; In response to a superpixel area of ​​an oilstone image one frame after a target frame image matching at least two superpixel areas in the target frame image, the relevant superpixel areas in the target frame image are merged; in response to a superpixel area of ​​an oilstone image one frame after a target frame image matching a superpixel area without a corresponding area in the target frame image, the superpixel area closest to the superpixel area in the next oilstone image is searched for and merged.

[0010] The effect is that by matching and merging superpixel areas, the difference in the number of superpixel areas can be effectively reduced, the matching accuracy can be improved, the computational complexity can be reduced, and the detection efficiency and accuracy can be improved.

[0011] Preferably, the temperature anomaly characteristic value includes: Taking any oilstone image as the target frame image and any superpixel area in the target frame image as the salient area, the ratio between the grayscale mean of the pixels in the salient area of ​​the target frame image and the grayscale mean of all the pixels in the target frame image is calculated, and the absolute difference between 1 and the ratio is taken as the deviation degree of the grayscale mean; the ratio between the maximum value of the grayscale gradient of the pixels in the salient area of ​​the target frame image and the maximum value of the grayscale gradient of all the pixels in the target frame image is calculated to obtain the maximum gradient ratio; the ratio of the standard deviation of the grayscale values ​​of all the pixels in the salient area of ​​the target frame image to the grayscale mean is taken as the coefficient of variation; the product of the deviation degree of the grayscale mean, the coefficient of variation and the maximum gradient ratio is taken as the temperature anomaly characteristic value of the salient area of ​​the target frame image.

[0012] The effect is that by comprehensively considering the grayscale mean deviation, coefficient of variation and maximum gradient ratio of the significant area, the temperature anomaly characteristics of the significant area can be accurately quantified. By calculating the deviation of the grayscale mean, the grayscale difference between the significant area and the overall image is reflected; the coefficient of variation measures the discrete degree of the grayscale value in the area, revealing the uniformity of the temperature distribution; the maximum gradient ratio reflects the severity of the temperature change. Multiplying these three to obtain the temperature anomaly characteristic value can effectively highlight the potential temperature anomaly area and provide a key basis for subsequent analysis.

[0013] Preferably, the correlation coefficient includes: According to the correspondence of superpixel regions, the correspondence of the same superpixel region in different frames of oilstone images is determined, any frame of oilstone image is taken as the target frame image, any superpixel region in the target frame image is taken as the significant region, the grayscale mean of the significant region in the target frame image is obtained, and the grayscale mean sequence of the significant region in the whole cooling process is obtained; Traverse all superpixel regions and use the Pearson correlation coefficient to calculate the correlation coefficient between the grayscale mean sequence of the salient region and the grayscale mean sequence of other superpixel regions.

[0014] The effect is: by determining the correspondence between superpixel areas in different frames, the changes of each area during the cooling process can be accurately tracked, and the grayscale mean sequence of the significant area during the entire cooling process can be calculated to reflect the temperature change trend of the area. The Pearson correlation coefficient is used to calculate the correlation coefficient of the grayscale mean sequence of the significant area and other areas, which can quantify the similarity of temperature changes between different areas, and then evaluate the uniformity and potential anomalies of pore distribution. Areas with low correlation coefficients may indicate abnormal pore distribution because the temperature change trend of these areas is inconsistent with other areas, which helps to identify areas where internal defects may exist.

[0015] Preferably, the comprehensive correlation index includes: The absolute value of the correlation coefficient is used as the temperature change trend during the cooling process, any oilstone image is used as the target frame image, any superpixel area in the target frame image is used as the significant area, the minimum value of the Euclidean distance between the central pixel points of the significant area and other superpixel areas is used as the distance weight, and the average of the temperature anomaly feature values ​​of other superpixel areas in each frame image is used as the temperature anomaly weight; The product of the temperature change trend, the negative of the distance weight and the temperature anomaly weight is taken as the contribution value of the correlation coefficient of other superpixel areas to the significant area, and the sum of the contribution values ​​of the correlation coefficients of all superpixel areas is taken as the comprehensive correlation index of the significant area.

[0016] The effect is: by calculating the correlation coefficient of the significant area with other areas, combining the distance weight and the temperature anomaly weight for weighting, the correlation of the temperature change trends between different areas can be accurately quantified. The product of the temperature change trend, the negative of the distance weight and the temperature anomaly weight is used as the contribution value to highlight the potential temperature anomaly area. The contribution values ​​of the correlation coefficients of all super-pixel areas are summed up to obtain the comprehensive correlation index of the significant area, which is helpful to comprehensively evaluate the consistency of the temperature change trend of the significant area with other areas. Through the comprehensive correlation index, areas with possible internal defects can be more accurately identified, thereby improving the accuracy of static pressing quality detection.

[0017] Preferably, the probability value of the occurrence of pore anomaly includes: Taking any oilstone image as the target frame image, taking any superpixel area in the target frame image as the significant area, obtaining the maximum value of the temperature anomaly characteristic value of all superpixel areas in the target frame image, calculating the ratio between the comprehensive correlation index of the significant area and the maximum value of the comprehensive correlation index of all superpixel areas in the target frame image, and obtaining the relative comprehensive correlation coefficient; The product of the maximum value of the temperature anomaly characteristic value and the relative comprehensive correlation coefficient is normalized to obtain the possibility value of the occurrence of pore anomaly in the significant area.

[0018] The effect is that by combining the maximum value of the temperature anomaly characteristic value and the relative comprehensive correlation coefficient, the normalization process can transform the complex index into an intuitive possibility value, accurately quantify the probability of pore anomaly inside the significant area, and normalize the result to a fixed range, usually The higher the value, the greater the possibility of pore anomaly. Taking temperature anomaly and correlation into consideration, the pore anomaly area is highlighted to improve detection accuracy.

[0019] Preferably, the determining whether there is abnormal pore distribution inside the oilstone includes: In response to the possibility value of pore anomaly occurring inside the superpixel area being greater than or equal to the preset pore anomaly threshold, pore distribution anomaly occurs inside the area and the control system marks it; otherwise, if it is less than the preset pore anomaly threshold, the pore distribution inside the area is uniform and meets the standard.

[0020] Preferably, preprocessing the infrared image set includes: A filtering algorithm is used to denoise each frame of the oilstone image in the infrared image set, the infrared image set is aligned, and the background of each frame of the oilstone image in the infrared image set is normalized to reduce the impact of background differences on defect detection.

[0021] Preferably, the background is normalized, including: The original infrared thermal images of multiple frames of the oilstone surface before heating are obtained, and the average value of the original infrared thermal images of multiple frames of the oilstone surface is calculated to obtain a reference background image. After the oilstone undergoes a high-temperature sintering process, an infrared detector is used to obtain a set of infrared images of the oilstone surface in the cooling stage at a fixed frequency. The reference background image is subtracted from the infrared image sequence after heating to obtain the actual infrared radiation rise value of the oilstone surface caused by long pulse excitation, thereby completing the background normalization processing.

[0022] The effect is that background normalization can effectively eliminate the influence of uneven initial thermal radiation distribution of oilstone, reduce the interference of background difference on subsequent defect detection, and improve detection accuracy. After background normalization, the actual infrared radiation rise value caused by long pulse excitation on the oilstone surface can be obtained, which more accurately reflects the thermal radiation change of oilstone during the cooling process, and is helpful for the subsequent analysis of abnormal pore distribution inside the oilstone.

[0023] The present invention has the following effects: 1. The present invention uses superpixel segmentation and matching, combined with temperature anomaly characteristic values ​​and comprehensive correlation indicators, to accurately evaluate the possibility of pore anomalies in each superpixel area, effectively discover potential defects inside the oilstone, and ensure the quality of static pressing.

[0024] 2. The present invention quickly establishes the correspondence between superpixel areas in adjacent frame images through superpixel segmentation, matching and correlation analysis, highlights potential temperature anomaly areas, improves detection efficiency, and comprehensively considers multiple factors to improve detection accuracy.

[0025] 3. The present invention analyzes the temperature change trend and correlation, combines the distance weight and the temperature anomaly weight for weighted fusion, makes the detection result more convincing and reliable, and provides a stable basis for the quality detection of oilstone static pressure molding. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a method flow chart of steps S1 to S4 in a method for detecting quality of oilstone static pressing molding based on image processing in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0028] Reference Figure 1 A method for detecting the quality of oilstone static pressure forming based on image processing includes steps S1 to S4, which are as follows: S1: Obtain a set of infrared images of the oilstone sintering cooling stage and perform preprocessing.

[0029] Preprocess the infrared image collection, including: A filtering algorithm is used to denoise each frame of the oilstone image in the infrared image set, the infrared image set is aligned, and the background of each frame of the oilstone image in the infrared image set is normalized to reduce the impact of background differences on defect detection.

[0030] In order to eliminate the influence of the movement of the oilstone sample during the sintering and cooling process, it is necessary to align the infrared image set. The feature point matching algorithm (such as SIFT algorithm) is used to first detect the feature points in each frame image, and then transform and align the image according to the matching relationship of the feature points, so as to ensure the consistency of the position and posture of the oilstone in the infrared image set. The feature point matching algorithm is a well-known technology in the field and will not be described in detail.

[0031] The background is normalized, including: The original infrared thermal images of multiple frames of the oilstone surface before heating are obtained, and the average value of the original infrared thermal images of multiple frames of the oilstone surface is calculated to obtain a reference background image. After the oilstone undergoes a high-temperature sintering process, an infrared detector is used to obtain a set of infrared images of the oilstone surface in the cooling stage at a fixed frequency. The reference background image is subtracted from the infrared image sequence after heating to obtain the actual infrared radiation rise value of the oilstone surface caused by long pulse excitation, thereby completing the background normalization processing.

[0032] The purpose of acquiring a reference background image is to obtain a stable and representative background thermal radiation distribution.

[0033] S2: Perform superpixel segmentation on each oilstone image in the infrared image set to obtain superpixel regions, perform superpixel region matching on two adjacent oilstone images in the infrared image set in turn to establish a corresponding relationship, and calculate the temperature anomaly characteristic value of each superpixel region in each oilstone image during the cooling process according to the matching results.

[0034] Superpixel areas, including: Obtain the grayscale values ​​of all pixels in each frame of the oilstone image, and use the Sobel operator to calculate the gradient amplitude of each pixel; Based on the grayscale value, gradient value and position coordinates of the pixel points in each frame of the oilstone image, the SLIC superpixel segmentation algorithm is used to perform superpixel segmentation on the infrared image set to obtain the superpixel area.

[0035] Exemplarily, the number of superpixel regions is set to 100, which can be adjusted appropriately according to the image size.

[0036] Perform superpixel region matching, including: Taking any oilstone image as the target frame image, obtaining the coordinates of the central pixel point of the superpixel region in the target frame image, and calculating the Euclidean distance between the central pixel point of the superpixel region in the target frame image and the next oilstone image to match the corresponding area; In response to a superpixel area of ​​an oilstone image one frame after a target frame image matching at least two superpixel areas in the target frame image, the relevant superpixel areas in the target frame image are merged; in response to a superpixel area of ​​an oilstone image one frame after a target frame image matching a superpixel area without a corresponding area in the target frame image, the superpixel area closest to the superpixel area in the next oilstone image is searched for and merged.

[0037] Temperature anomaly characteristic values, including: Taking any oilstone image as the target frame image and any superpixel area in the target frame image as the salient area, the ratio between the grayscale mean of the pixels in the salient area of ​​the target frame image and the grayscale mean of all the pixels in the target frame image is calculated, and the absolute difference between 1 and the ratio is taken as the deviation degree of the grayscale mean; the ratio between the maximum value of the grayscale gradient of the pixels in the salient area of ​​the target frame image and the maximum value of the grayscale gradient of all the pixels in the target frame image is calculated to obtain the maximum gradient ratio; the ratio of the standard deviation of the grayscale values ​​of all the pixels in the salient area of ​​the target frame image to the grayscale mean is taken as the coefficient of variation; the product of the deviation degree of the grayscale mean, the coefficient of variation and the maximum gradient ratio is taken as the temperature anomaly characteristic value of the salient area of ​​the target frame image.

[0038] It should be noted that in the actual physical process, the temperature distribution on the surface of the oilstone is affected by many factors. The temperature anomaly in a certain area is usually not caused by a single factor, but the result of multiple factors working together. Therefore, the deviation degree of the grayscale mean, the coefficient of variation, and the maximum gradient ratio comprehensively consider the deviation degree of the grayscale mean, the uniformity of the temperature distribution in the area, and the severity of the temperature gradient, which can more comprehensively reflect the possibility of temperature anomaly in the area. Only when all three aspects show abnormalities, the area is more likely to have real temperature anomalies.

[0039] Specifically, the temperature anomaly characteristic value satisfies the following relationship: ; In the formula, Indicates Frame oilstone image The temperature anomaly characteristic value of the super-pixel area, Indicates Frame oilstone image The grayscale mean of the superpixel area, Indicates The grayscale mean of all pixels in the frame oilstone image, Indicates Frame oilstone image The coefficient of variation of the grayscale values ​​of all pixels in the superpixel area is Indicates Frame oilstone image The maximum value of the grayscale gradient of the pixels in the superpixel area, Indicates The maximum value of the grayscale gradient of all pixels in the frame oilstone image.

[0040] That is to say, For the Frame oilstone image The ratio of the grayscale mean of the superpixel area to the grayscale mean of the entire oilstone surface area, The closer it is to 1, the The closer the temperature of a superpixel area is to the average temperature of the entire oilstone surface area, the smaller the possibility of abnormality. It may be caused by the difference in heat conduction between the area and the whole body. For example, the presence of internal pores will change the heat conduction path, causing the local temperature to deviate from the overall average temperature.

[0041] The coefficient of variation reflects the degree of dispersion of the grayscale values ​​of pixels in the area. The larger the value, the more uneven the temperature distribution in the area, and a large temperature gradient may appear, reflecting uneven heat dissipation in the area and a greater possibility of abnormal internal pore distribution. Related to the inhomogeneity of the microstructure within the region. If the pore distribution within the region is uneven or there are differences in material properties, it will lead to uneven temperature distribution, which will increase the coefficient of variation.

[0042] It is Frame oilstone image The ratio of the maximum grayscale gradient of the pixels in the superpixel area to the maximum grayscale gradient of the pixels in the entire oilstone surface area. The larger the grayscale gradient of the pixels in this area, the greater the temperature gradient, the worse the heat dissipation effect, and the greater the possibility of abnormal internal pore distribution. Related to the heat flow distribution in the area. Sharp temperature gradients may mean that heat flow is blocked or there are abnormal heat sources / sinks, which are related to abnormal pore distribution or other internal defects.

[0043] S3: Based on the analysis of the temperature change trend of each superpixel area in different frames of oilstone images, the correlation coefficient between each superpixel area and other superpixel areas is calculated, and the distance weight and temperature anomaly weight are introduced for weighted fusion to obtain a comprehensive correlation index. Combined with the temperature anomaly characteristic value, the possibility of whether there is pore anomaly in each superpixel area is evaluated.

[0044] Correlation coefficients, including: According to the correspondence of superpixel regions, the correspondence of the same superpixel region in different frames of oilstone images is determined, any frame of oilstone image is taken as the target frame image, any superpixel region in the target frame image is taken as the significant region, the grayscale mean of the significant region in the target frame image is obtained, and the grayscale mean sequence of the significant region in the whole cooling process is obtained; Traverse all superpixel regions and use the Pearson correlation coefficient to calculate the correlation coefficient between the grayscale mean sequence of the salient region and the grayscale mean sequence of other superpixel regions.

[0045] Further analysis shows that the heat dissipation effect at the edge of the oilstone is better than that in the central area. Therefore, the correlation between the temperature changes of different superpixel areas on the surface of the oilstone will vary with the distance. If there is an abnormality in the pore distribution inside a superpixel area, the correlation between the temperature changes of the superpixel area and other superpixel areas will decrease. In order to avoid the influence of distance, the correlation value of the grayscale change sequence between regions is weighted.

[0046] The closer the distance is, the higher the consistency of the temperature change trend is, and the higher the confidence of the correlation value is. Conversely, the farther the distance is, the lower the consistency of the temperature change is, and the lower the confidence of the correlation value is. Therefore, when calculating the correlation value, the distance weight is introduced to give a lower weight to the correlation value of the farther distance area.

[0047] If there is an abnormality in the internal pore distribution of a region, its heat dissipation effect will be worse, and the mean value of the temperature anomaly characteristic value will be larger. In this case, the correlation between the temperature change trend of this region and other regions may decrease. Therefore, when calculating the correlation value, the mean value of the temperature anomaly characteristic value is introduced as a weight, and a lower weight is given to the correlation value of the region with a larger mean value of the temperature anomaly characteristic value.

[0048] The specific weighting steps are as follows: Comprehensive relevance indicators, including: The absolute value of the correlation coefficient is used as the temperature change trend during the cooling process, any oilstone image is used as the target frame image, any superpixel area in the target frame image is used as the significant area, the minimum value of the Euclidean distance between the central pixel points of the significant area and other superpixel areas is used as the distance weight, and the average of the temperature anomaly feature values ​​of other superpixel areas in each frame image is used as the temperature anomaly weight; The product of the temperature change trend, the negative of the distance weight and the temperature anomaly weight is taken as the contribution value of the correlation coefficient of other superpixel areas to the significant area, and the sum of the contribution values ​​of the correlation coefficients of all superpixel areas is taken as the comprehensive correlation index of the significant area.

[0049] By weighting the correlation values ​​of the grayscale change sequences between regions, the correlation of temperature change trends between different regions can be measured more accurately, avoiding the influence of distance and pore anomalies on the correlation analysis, thereby more accurately evaluating the possibility of abnormal pore distribution within each superpixel region.

[0050] It should be noted that since the heat dissipation effect at the edge of the oilstone is usually better than that in the center, the correlation of temperature changes between different areas will change with distance. The closer the distance is, the higher the consistency of the temperature change trend is, and the higher the confidence of the correlation value is; conversely, the farther the distance is, the lower the consistency of the temperature change is, and the lower the confidence of the correlation value is. Therefore, when calculating the comprehensive correlation index, the distance weight is introduced. To correct the confidence of the correlation value.

[0051] Further analysis shows that if there is an abnormality in the internal pore distribution of a superpixel area, its heat dissipation effect will be worse, and the average temperature anomaly characteristic value will be larger. In this case, the correlation between the temperature change trend of this area and other areas may decrease. Therefore, when calculating the comprehensive correlation index, the mean of the temperature anomaly characteristic value is introduced as a weight to correct the correlation value. The larger the mean of the temperature anomaly characteristic value, the smaller the correlation contribution value.

[0052] Specifically, the comprehensive correlation index satisfies the following relationship: ; In the formula, Indicates The comprehensive correlation index of superpixel regions, represents the number of superpixel regions, Indicates The superpixel region is The Pearson correlation coefficient value between the grayscale mean sequences of superpixel regions is Indicates The superpixel region is The Euclidean distance of the center pixel of the superpixel region, Indicates The average value of the temperature anomaly feature value in all frame images of the super pixel area, Represents the minimum function.

[0053] That is to say, The larger the value, the greater the correlation between the temperature trends of the two regions during the cooling process; Indicates the first The superpixel region is The minimum value of the Euclidean distance between the central pixels of the superpixel areas. The closer the distance between the two areas, the higher the confidence of the correlation value. Conversely, the farther the distance between the two areas, the lower the consistency of their temperature changes and the lower the confidence of the correlation value.

[0054] The larger the value, the The greater the possibility of abnormal pore distribution inside the superpixel area, the worse the heat dissipation effect, and the greater the possibility of abnormal internal pore distribution. The lower the confidence level of the grayscale sequence correlation value of a region.

[0055] The probability value of the occurrence of porosity anomaly includes: Taking any oilstone image as the target frame image, taking any superpixel area in the target frame image as the significant area, obtaining the maximum value of the temperature anomaly characteristic value of all superpixel areas in the target frame image, calculating the ratio between the comprehensive correlation index of the significant area and the maximum value of the comprehensive correlation index of all superpixel areas in the target frame image, and obtaining the relative comprehensive correlation coefficient; The product of the maximum value of the temperature anomaly characteristic value and the relative comprehensive correlation coefficient is normalized to obtain the possibility value of the occurrence of pore anomaly in the significant area.

[0056] Specifically, the likelihood value satisfies the following relationship: ; In the formula, Indicates The probability value of the pore anomaly appearing inside the superpixel area, represents the normalization function, represents the maximum value function, Indicates Frame oilstone image The temperature anomaly characteristic value of the super-pixel area, Indicates The comprehensive correlation index of superpixel regions, Indicates the maximum value of the comprehensive correlation index in the superpixel area.

[0057] In other words, the upper limit of the temperature anomaly characteristic value of all superpixel regions in the target frame image is obtained, with the purpose of identifying the superpixel region with the most abnormal heat dissipation in the target frame image and highlighting the extreme situation of potential uneven pore distribution. It represents the highest level of the temperature anomaly characteristic value of all superpixel regions in the current frame image, which is a key indicator for measuring the abnormal heat dissipation situation and can be used for subsequent analysis of the possibility of abnormal pore distribution inside the oilstone.

[0058] Reflects the The correlation between the temperature change trend of a superpixel area and other areas is the relative position in the whole. The smaller the value, the more inconsistent the temperature change trend of this area is with other areas, which may cause abnormal heat dissipation mode due to abnormal pores.

[0059] S4: According to the probability value of pores appearing in each superpixel area, determine whether there is abnormal pore distribution inside the oilstone, and complete the quality inspection goal of oilstone static pressing.

[0060] In response to the possibility value of pore anomaly occurring inside the superpixel area being greater than or equal to the preset pore anomaly threshold, pore distribution anomaly occurs inside the area and the control system marks it; otherwise, if it is less than the preset pore anomaly threshold, the pore distribution inside the area is uniform and meets the standard.

[0061] Exemplarily, the preset porosity anomaly threshold is 0.6, and implementers can adjust it according to the provision situation.

[0062] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these modifications and improvements all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for detecting the quality of oilstone static pressure forming based on image processing, characterized in that: include: Obtain a set of infrared images of the oilstone sintering cooling stage and perform preprocessing; Perform superpixel segmentation on each oilstone image in the infrared image set to obtain superpixel regions, perform superpixel region matching on two adjacent oilstone images in the infrared image set in turn to establish a corresponding relationship, and calculate temperature anomaly characteristic values ​​of each superpixel region in each oilstone image during the cooling process according to the matching results; Based on the analysis of the temperature change trend of each superpixel area in different frames of oilstone images, the correlation coefficient between each superpixel area and other superpixel areas is calculated, and the distance weight and temperature anomaly weight are introduced for weighted fusion to obtain a comprehensive correlation index. Combined with the temperature anomaly characteristic value, the possibility of pore anomaly in each superpixel area is evaluated. According to the probability value of pores appearing in each superpixel area, it is determined whether there is abnormal pore distribution inside the oilstone, and the quality inspection goal of oilstone static pressing forming is completed.

2. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The super pixel area comprises: Obtain the grayscale value of all pixels in each frame of the oilstone image, and use the Sobel operator to calculate the gradient amplitude of each pixel; Based on the grayscale value, gradient value and position coordinates of the pixel points in each frame of the oilstone image, the SLIC superpixel segmentation algorithm is used to perform superpixel segmentation on the infrared image set to obtain the superpixel area.

3. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The step of sequentially performing superpixel region matching on two adjacent oilstone images in the infrared image set includes: Taking any oil stone image as the target frame image, the coordinates of the central pixel point of the superpixel area in the target frame image are obtained, and the Euclidean distance between the central pixel point of the superpixel area in the target frame image and the next oil stone image is calculated to match the corresponding area; in response to the superpixel area of ​​the oil stone image in the next frame after the target frame image matching at least two superpixel areas in the target frame image, the relevant superpixel areas in the target frame image are merged; in response to the superpixel area of ​​the oil stone image in the next frame after the target frame image matching a superpixel area without a corresponding area in the target frame image, the superpixel area closest to the superpixel area in the next frame of the oil stone image is searched for and merged.

4. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The temperature anomaly characteristic value includes: Taking any oilstone image as the target frame image and any superpixel area in the target frame image as the salient area, the ratio between the grayscale mean of the pixels in the salient area of ​​the target frame image and the grayscale mean of all the pixels in the target frame image is calculated, and the absolute difference between 1 and the ratio is taken as the deviation degree of the grayscale mean; the ratio between the maximum value of the grayscale gradient of the pixels in the salient area of ​​the target frame image and the maximum value of the grayscale gradient of all the pixels in the target frame image is calculated to obtain the maximum gradient ratio; the ratio of the standard deviation of the grayscale values ​​of all the pixels in the salient area of ​​the target frame image to the grayscale mean is taken as the coefficient of variation; the product of the deviation degree of the grayscale mean, the coefficient of variation and the maximum gradient ratio is taken as the temperature anomaly characteristic value of the salient area of ​​the target frame image.

5. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The correlation coefficient includes: According to the correspondence of superpixel regions, the correspondence of the same superpixel region in different frames of oilstone images is determined, any frame of oilstone image is taken as the target frame image, any superpixel region in the target frame image is taken as the significant region, the grayscale mean of the significant region in the target frame image is obtained, and the grayscale mean sequence of the significant region in the whole cooling process is obtained; Traverse all superpixel regions and use the Pearson correlation coefficient to calculate the correlation coefficient between the grayscale mean sequence of the salient region and the grayscale mean sequence of other superpixel regions.

6. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The comprehensive correlation indicators include: The absolute value of the correlation coefficient is used as the temperature change trend during the cooling process, any oilstone image is used as the target frame image, any superpixel area in the target frame image is used as the significant area, the minimum value of the Euclidean distance between the central pixel points of the significant area and other superpixel areas is used as the distance weight, and the average of the temperature anomaly feature values ​​of other superpixel areas in each frame image is used as the temperature anomaly weight; The product of the temperature change trend, the negative of the distance weight and the temperature anomaly weight is taken as the contribution value of the correlation coefficient of other superpixel areas to the significant area, and the sum of the contribution values ​​of the correlation coefficients of all superpixel areas is taken as the comprehensive correlation index of the significant area.

7. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The probability value of the occurrence of pore anomaly includes: Taking any oilstone image as the target frame image, taking any superpixel area in the target frame image as the significant area, obtaining the maximum value of the temperature anomaly characteristic value of all superpixel areas in the target frame image, calculating the ratio between the comprehensive correlation index of the significant area and the maximum value of the comprehensive correlation index of all superpixel areas in the target frame image, and obtaining the relative comprehensive correlation coefficient; The product of the maximum value of the temperature anomaly characteristic value and the relative comprehensive correlation coefficient is normalized to obtain the possibility value of the occurrence of pore anomaly in the significant area.

8. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: The determination of whether there is abnormal pore distribution inside the oilstone includes: In response to the possibility value of pore anomaly occurring inside the superpixel area being greater than or equal to the preset pore anomaly threshold, pore distribution anomaly occurs inside the area and the control system marks it; otherwise, if it is less than the preset pore anomaly threshold, the pore distribution inside the area is uniform and meets the standard.

9. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 1 is characterized in that: Preprocessing the infrared image set includes: A filtering algorithm is used to denoise each frame of the oilstone image in the infrared image set, the infrared image set is aligned, and the background of each frame of the oilstone image in the infrared image set is normalized to reduce the impact of background differences on defect detection.

10. The method for detecting the quality of oilstone static pressing molding based on image processing according to claim 9 is characterized in that: The background is normalized, including: The original infrared thermal images of multiple frames of the oilstone surface before heating are obtained, and the average value of the original infrared thermal images of multiple frames of the oilstone surface is calculated to obtain a reference background image. After the oilstone undergoes a high-temperature sintering process, an infrared detector is used to obtain a set of infrared images of the oilstone surface in the cooling stage at a fixed frequency. The reference background image is subtracted from the infrared image sequence after heating to obtain the actual infrared radiation rise value of the oilstone surface caused by long pulse excitation, thereby completing the background normalization processing.

Images