Aluminum foil detection system and method based on computer vision

Abstract

The invention discloses an operation monitoring system and method for waste gas treatment equipment. The method comprises the following steps: acquiring a shot image of an aluminum foil defect test chart; aluminum foil defects are analyzed to obtain image similarity values; comparing the image similarity value with an image similarity threshold value, judging whether the test image is qualified or not according to a comparison result, and generating a test non-passing signal and a test passing signal; analyzing association between the unqualified test image and equipment illumination based on the test failing signal to obtain an illumination association value; comparing the illumination association value with an illumination association threshold value, judging whether the test image is unqualified and is associated with an illumination condition according to a comparison result, and if so, generating an illumination association signal; based on the test passing signal, the aluminum foil defect points are analyzed, and a detection defect value is obtained; comparing the detection defect value with a detection defect threshold value, and judging whether the produced aluminum foil is qualified or not according to a comparison result; the aluminum foil surface defect identification precision and detection accuracy of the system can be improved.

Description

Technical Field

[0002] The invention belongs to the technical field of industrial automation, and in particular relates to an aluminum foil detection system and method based on computer vision. Background Art

[0004] As an important industrial material, the production process of aluminum foil is gradually shifting towards intelligence and automation. The computer vision-based detection system is an important product of this trend. However, various flaws and defects may occur in the production, processing and transportation of aluminum foil, such as scratches, bubbles, stains, creases, etc. These flaws and defects will affect the quality and performance of the aluminum foil, and thus affect the quality and safety of the product. When the computer vision-based detection system detects aluminum foil, due to the strong reflectivity of the aluminum foil surface, the image of the aluminum foil surface is often unclear, and the defects on the aluminum foil surface cannot be accurately identified.

[0005] The present invention determines whether the aluminum foil surface image obtained by a computer vision-based detection system is clear, and whether the lighting conditions of the equipment cause reflections or shadows on the aluminum foil surface under the current environment; if the details of the aluminum foil surface image obtained by the equipment are not clear, the lighting conditions are optimized and adjusted; the clarity of the obtained aluminum foil surface image is improved, and the detection accuracy of the equipment is improved. Summary of the invention

[0007] The object of the present invention is to provide an operation monitoring system and method for exhaust gas treatment equipment to solve at least one of the above-mentioned problems in the prior art.

[0008] First,

[0009] Step 1: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil on the image, analyze the defects on the aluminum foil, and obtain an image similarity value;

[0010] Step 2: Compare the image similarity value with the image similarity threshold, and determine whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated;

[0011] Step 3: Based on the test failure signal, analyze the correlation between the test image failure and the device lighting condition to obtain the lighting correlation value;

[0012] Step 4: Compare the illumination correlation value with the illumination correlation threshold value, and determine whether the failure of the test image is related to the illumination condition according to the comparison result. If so, generate an illumination correlation signal and adjust the illumination condition until the test image passes;

[0013] Step 5: Based on the test pass signal, the aluminum foil is inspected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is determined based on the comparison result.

[0014] Second,

[0015] Test parameter acquisition module: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil in the image, analyze the defects on the aluminum foil, and obtain the image similarity value;

[0016] Test parameter analysis module: compares the image similarity value with the image similarity threshold, and determines whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated;

[0017] Correlation parameter acquisition module: Based on the test failure signal, the correlation between the test image failure and the device lighting conditions is analyzed to obtain the lighting correlation value;

[0018] Correlation parameter analysis module: compares the illumination correlation value with the illumination correlation threshold, and determines whether the failure of the test image is related to the illumination condition based on the comparison result. If so, generates an illumination correlation signal and adjusts the illumination condition until the test image passes;

[0019] Qualified detection module: Based on the test pass signal, the aluminum foil is detected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is determined based on the comparison result.

[0020] Beneficial effects of the present invention:

[0021] 1. Analyze the aluminum foil surface image obtained by the aluminum foil detection system based on computer vision, determine the defect points of the aluminum foil surface image, analyze the number and clarity of the defect points obtained, and determine whether the image of the aluminum foil surface obtained by the system is clear enough; if the clarity meets the detection conditions, detect the aluminum foil; if the clarity does not meet the detection conditions, analyze the lighting conditions of the detection equipment; the clarity of the obtained aluminum foil surface can improve the accuracy of the equipment's aluminum foil detection and improve the accuracy of aluminum foil detection.

[0022] 2. When the image clarity of the aluminum foil surface obtained by the computer vision-based aluminum foil detection system does not meet the detection conditions, the lighting conditions of the detection equipment are analyzed to determine whether the failure of the obtained aluminum foil surface image clarity to meet the requirements is related to the lighting conditions of the equipment; if there is a correlation, the lighting conditions of the detection equipment are adjusted and optimized to obtain a clearer image.

[0023] 3. When the aluminum foil detection system based on computer vision detects defects on the surface of aluminum foil, because the surface of the aluminum foil is smooth and reflects light more obviously, inappropriate lighting conditions may cause the surface of the aluminum foil to reflect light, or make some detailed defects on the surface of the aluminum foil unable to be identified, thus reducing the recognition accuracy. Analyzing the impact of the equipment lighting conditions on the acquired image and adjusting the lighting conditions according to the analysis results will help improve the detection accuracy of the aluminum foil detection system based on computer vision. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0026] Figure 1 This is a flow chart of obtaining an image similarity value by a computer vision-based aluminum foil detection method provided in Embodiment 1 of the present invention;

[0027] Figure 2 This is a flow chart of obtaining illumination correlation values ​​using a computer vision-based aluminum foil detection method provided in Embodiment 2 of the present invention;

[0028] Figure 3 This is a flow chart of obtaining a detection defect value by an aluminum foil detection system based on computer vision provided in Embodiment 3 of the present invention;

[0029] Figure 4 It is a module schematic diagram of an aluminum foil detection system based on computer vision provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described 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, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] Embodiment 1

[0033] like Figure 1 As shown, an aluminum foil detection method based on computer vision provided by an embodiment of the present invention specifically includes the following steps:

[0034] Step 1: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil on the image, analyze the defects on the aluminum foil, and obtain an image similarity value;

[0035] It should be noted that the aluminum foil defect test chart is used to evaluate the performance and accuracy of the detection system by simulating the surface conditions of aluminum foil in actual production; including but not limited to scratches, stains and other defects on the aluminum foil;

[0036] Images of various aluminum foil defect test patterns are acquired through a high-resolution industrial camera and marked as test images;

[0037] Obtaining defect data in the test image and performing data analysis on the defect data; wherein the defect data in the test image includes but is not limited to the edge length, shape and size of the defect, and the number of defects;

[0038] Based on any defect point in the aluminum foil defect test chart;

[0039] Obtain the defect perimeter, defect area, defect major axis and defect minor axis of the defect point;

[0040] It should be noted that the defect major axis of a defect point is the maximum value of the defect point length, and the defect minor axis is the maximum value of the defect point width;

[0041] In some embodiments, the defect area is separated from the background by an image segmentation algorithm; on the segmented defect area, a contour extraction algorithm, such as a Canny edge detector, is used to obtain the contour of the defect, and the perimeter of the defect and the major axis and minor axis of the defect are calculated based on the extracted contour; the pixels of the segmented defect area are counted, and then the defect area is obtained based on the actual area size corresponding to each pixel;

[0042] The defect major axis and defect minor axis are processed as ratios to obtain the image aspect ratio;

[0043] The defect perimeter and defect area are processed and the formula is used , the image density JM is obtained; where P represents the defect perimeter and A represents the defect area;

[0044] It should be noted that image density indicates how compact the image is. When the image is circular, the image density reaches its maximum value, which is 1. For other shapes, the density is usually less than 1. The more irregular the shape, the elongated or flat the image is, and the smaller the density value is. A shape with a smaller density means that its area is smaller under a given perimeter, or its perimeter is larger under a given area.

[0045] The image aspect ratio of the test image is compared with the standard image aspect ratio of the aluminum foil defect test image to obtain the aspect difference ratio;

[0046] The image compactness of the test image is compared with the standard image compactness of the aluminum foil defect test image to obtain a compactness difference ratio;

[0047] The obtained length-width difference ratio and compactness difference ratio are multiplied to obtain the defect deformation value;

[0048] Compare the defect deformation value with the defect deformation threshold. If the defect deformation value is greater than the defect deformation threshold, it means that the defect point in the test image is more deformed than the defect point in the aluminum foil defect test image. Then the acquired defect point imaging is unqualified and marked as an unclear defect point. Otherwise, it is marked as a clear defect point.

[0049] Get the number of unclear defect points in the aluminum foil defect test image;

[0050] The number of unclear defect points is compared with the total number of defect points in the test image to obtain the defect clarity ratio, which is marked as QX;

[0051] Obtain the total number of defect points identified in the test image and the total number of defect points in the aluminum foil defect test image; perform ratio processing on the total number of defect points identified in the test image and the total number of defect points in the aluminum foil defect test image to obtain a defect recognition ratio, which is marked as SB;

[0052] The obtained defect clarity ratio QX and defect recognition ratio SB are processed and the formula is used , and obtain the image similarity value XS; where s1 and s2 are preset proportional coefficients;

[0053] It should be noted that the image similarity value indicates the similarity between the acquired test image and the aluminum foil defect test image; the larger the image similarity value, the higher the similarity between the test image and the aluminum foil defect test image, which means that the test image acquired by the high-resolution industrial camera is clearer;

[0054] Step 2: Compare the image similarity value with the image similarity threshold, and determine whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated;

[0055] The image similarity value is compared with the image similarity threshold. The comparison process is as follows:

[0056] If the image similarity value is greater than or equal to the image similarity threshold, a test pass signal is generated;

[0057] If the image similarity value is less than the image similarity threshold, a test failure signal is generated;

[0058] Based on the test passing signal, marking the test image as a passed test image;

[0059] Based on the test failure signal, marking the test image as a failed test image;

[0060] The technical solution of the embodiment of the present invention is: acquiring a test image through a high-resolution industrial camera, processing the defect points of the test image, analyzing the similarity between the defect points of the test image and the defect points of the aluminum foil defect test image; judging whether the device can clearly identify the defect points; and then judging whether the device can identify all the defect points by analyzing the number of defect points identified in the test image; the present invention analyzes the number of aluminum foil defect points identified by the device and the clarity of the identified defect points to judge whether the test image is qualified, and can find out whether the image recognition of the aluminum foil by the device is clear, and judge whether there is an error in the detection of the aluminum foil by the device.

[0061] Embodiment 2

[0062] like Figure 2 As shown, an aluminum foil detection method based on computer vision provided by an embodiment of the present invention specifically includes the following steps:

[0063] Step 3: Based on the test failure signal, analyze whether there is a correlation between the test image failure and the device lighting condition, and obtain the lighting correlation value;

[0064] It should be noted that the lighting conditions of the equipment include but are not limited to lighting angle and brightness;

[0065] Based on the test failure signal, analyze whether there is a correlation between the test image failure and the device lighting conditions;

[0066] In some embodiments, aluminum foil is a reflective material, and its surface may present different reflective properties due to the production process or processing process; when the brightness is too high, it is necessary to reduce the brightness of the light source to avoid overexposure caused by excessive reflection; when the brightness is too low, it is necessary to increase the brightness of the light source to ensure that the details of the aluminum foil surface can be clearly presented; when the lighting angle is not appropriate, it will cause uneven lighting; uneven lighting conditions of the detection system may cause differences in light intensity in different areas of the collected image, thereby affecting the accuracy of detail detection;

[0067] Get the areas of overly bright spots or stripes on the surface of the test image;

[0068] Mark the area where the spots or stripes on the surface of the test image are too bright as the reflective area; obtain the area of ​​the reflective area to obtain the reflective area;

[0069] Compare the reflective area with the total area of ​​the test image to obtain the reflective area ratio;

[0070] Obtain the number of unclear defect points in the reflective area, and perform ratio processing on it with the total number of unclear points to obtain the reflective defect ratio;

[0071] The reflective defect ratio and the reflective area ratio are processed as ratios to obtain the reflective correlation ratio;

[0072] Obtain the area of ​​the identified detail region and the area of ​​the unidentified detail region in the test image;

[0073] It should be noted that the detail area refers to an area containing a series of high-contrast details in the aluminum foil defect test image. The high-contrast details include but are not limited to fine lines, textures or patterns, etc.

[0074] In some embodiments, by imaging a high-contrast detail test image and analyzing the detail presentation of the high-contrast detail area in the imaging, it is determined whether the aluminum foil detection system device of the computer vision can detect the detail defects on the surface of the aluminum foil;

[0075] The number of unclear defect points in the unidentified detail area is obtained, and the ratio is processed with the total number of unclear points to obtain the detail defect ratio;

[0076] The reflection correlation ratio and detail defect ratio are weighted and summed to obtain the detail performance value, which is marked as XS;

[0077] Get the area of ​​the identified detail area;

[0078] The detail area in the test image is evenly divided into several sub-areas, and the area of ​​the detail area identified in the sub-area is obtained, which is marked as the sub-area identification area;

[0079] Sum and average the identified areas of all sub-regions to obtain the mean of the identified areas of the sub-regions;

[0080] Subtract the identified areas of all sub-regions from the mean of the identified areas of the sub-regions and take the absolute value to obtain the sub-region identified area deviation;

[0081] Analyze the deviation of sub-region identification area;

[0082] The sub-region identification area deviation is compared with the sub-region identification area deviation threshold, the number of sub-region identification area deviations greater than the sub-region identification area deviation threshold is obtained, and the number is compared with the total number of sub-region identification area deviations to obtain a sub-region deviation ratio, which is marked as FG;

[0083] It should be noted that the sub-area deviation ratio indicates that the device illumination is uneven, resulting in different levels of detail recognition of the aluminum foil test pattern surface between sub-areas;

[0084] The sub-region deviation ratio FG and detail performance value XS are processed and the formula is used , and obtain the illumination correlation value XJ; where s1 and s2 are preset proportional coefficients, and both are greater than zero;

[0085] It should be noted that the larger the illumination correlation value, the higher the correlation between the failure of the test image and the illumination;

[0086] Step 4: Compare the illumination correlation value with the illumination correlation threshold value, and determine whether the failure of the test image is related to the illumination condition according to the comparison result. If so, generate an illumination correlation signal and adjust the illumination condition until the test image passes;

[0087] The illumination correlation value is compared with the illumination correlation threshold. The comparison process is as follows:

[0088] If the illumination correlation value is greater than or equal to the illumination correlation threshold, a illumination correlation signal is generated;

[0089] If the illumination correlation value is less than the illumination correlation threshold, an illumination non-correlation signal is generated;

[0090] Analyze other reasons why the test image fails based on the illumination non-correlated signal;

[0091] Based on the illumination-related signal, the illumination condition is adjusted until the test image passes;

[0092] The technical solution of the embodiment of the present invention is: based on the test failure signal, whether there is a correlation between the test image failure and the equipment lighting conditions is analyzed to obtain a lighting correlation value; through the comparison result of the lighting correlation value and the lighting correlation threshold, it is judged whether the test image failure is associated with the lighting conditions, and if so, the lighting conditions are adjusted until the test image is qualified; the present invention analyzes the influence of the lighting conditions on the recognition of aluminum foil defects and adjusts the lighting conditions to avoid reflections during the aluminum foil detection process when the brightness is too high; to avoid the inability to recognize small defects of the aluminum foil when the brightness is too low; and to avoid the situation where the image acquisition of some areas is unclear due to uneven lighting.

[0093] Embodiment 3

[0094] like Figure 3 As shown, an aluminum foil detection method based on computer vision provided by an embodiment of the present invention specifically includes the following steps:

[0095] Step 5: Based on the test pass signal, the aluminum foil is inspected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is determined according to the comparison result;

[0096] Based on the test pass signal, the surface image of the aluminum foil is collected, and the collected image is transmitted to the image processing unit for preprocessing; the preprocessed image is transmitted to the defect recognition unit, and the defect recognition unit recognizes defects on the image, and the recognized defects include but are not limited to the number of defect points, the area of ​​defect points, etc.;

[0097] Obtain the number of defect points on the aluminum foil surface and the area of ​​each defect point; sum up the areas of all defect points to obtain the total defect area; compare the total defect area with the total area of ​​the aluminum foil surface to obtain the defect area ratio;

[0098] Compare the number of defect points on the aluminum foil surface with the defect point threshold to obtain the defect point value ratio;

[0099] The defect area ratio and the defect point value ratio are weighted and summed to obtain the detection defect value;

[0100] The detected defect value is compared with the detected defect threshold. The comparison process is as follows:

[0101] If the detection defect value is greater than or equal to the detection defect threshold, an aluminum foil unqualified signal is generated;

[0102] If the detected defect value is less than the detected defect threshold, a qualified aluminum foil signal is generated;

[0103] Based on the aluminum foil unqualified signal, an aluminum foil unqualified alarm is issued to prompt the administrator of the unqualified;

[0104] The technical solution of the embodiment of the present invention is: based on the test pass signal, the aluminum foil is detected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is judged according to the comparison result; the present invention judges whether the aluminum foil is qualified by analyzing the defect points on the surface of the aluminum foil; it can quickly and effectively judge whether the aluminum foil is qualified, and issue an alarm to prompt the administrator to stop the production of unqualified aluminum foil in time.

[0105] Embodiment 4

[0106] like Figure 4 As shown, an aluminum foil detection system based on computer vision provided by an embodiment of the present invention specifically includes the following modules:

[0107] Test parameter acquisition module: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil in the image, analyze the defects on the aluminum foil, and obtain the image similarity value;

[0108] Test parameter analysis module: compares the image similarity value with the image similarity threshold, and determines whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated;

[0109] Correlation parameter acquisition module: Based on the test failure signal, the correlation between the test image failure and the device lighting conditions is analyzed to obtain the lighting correlation value;

[0110] Correlation parameter analysis module: compares the illumination correlation value with the illumination correlation threshold, and determines whether the failure of the test image is related to the illumination condition based on the comparison result. If so, generates an illumination correlation signal and adjusts the illumination condition until the test image passes;

[0111] Qualified detection module: Based on the test pass signal, the aluminum foil is detected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is determined based on the comparison result.

[0112] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for detecting aluminum foil based on computer vision, characterized in that: The following steps are involved: Step 1: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil on the image, analyze the defects on the aluminum foil, and obtain an image similarity value; Step 2: Compare the image similarity value with the image similarity threshold, and determine whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated; Step 3: Based on the test failure signal, analyze the correlation between the test image failure and the device lighting condition to obtain the lighting correlation value; Step 4: Compare the illumination correlation value with the illumination correlation threshold value, and determine whether the failure of the test image is associated with the illumination condition according to the comparison result. If so, generate an illumination correlation signal and adjust the illumination condition. Step 5: Based on the test pass signal, the aluminum foil is inspected, and the defect points on the aluminum foil are analyzed to obtain the detection defect value; the detection defect value is compared with the detection defect threshold, and whether the produced aluminum foil is qualified is determined based on the comparison result.

2. The aluminum foil detection method based on computer vision according to claim 1, characterized in that: The method for obtaining the image similarity value is as follows: Perform data analysis on the number of unclear defect points to obtain the defect clarity ratio QX; Perform data processing on the total number of defect points identified in the test image to obtain a defect recognition ratio SB; The obtained defect clarity ratio QX and defect recognition ratio SB are processed and the formula is used , and obtain the image similarity value XS; where s1 and s2 are preset proportional coefficients.

3. The aluminum foil detection method based on computer vision according to claim 2, characterized in that: The defect clarity ratio is obtained as follows: Get the number of unclear defect points in the aluminum foil defect test image; The number of unclear defect points is compared with the total number of defect points in the test image to obtain the defect clarity ratio, which is marked as QX.

4. The aluminum foil detection method based on computer vision according to claim 3, characterized in that: The method for obtaining the unclear defect points is as follows: Obtain the defect perimeter, defect area, defect major axis and defect minor axis of the defect point; The defect major axis and defect minor axis are processed as ratios to obtain the image aspect ratio; The defect perimeter and defect area are processed and the formula is used , get the image compactness JM; Where P represents the defect perimeter and A represents the defect area; The image aspect ratio of the test image is compared with the standard image aspect ratio of the aluminum foil defect test image to obtain the aspect difference ratio; The image compactness of the test image is compared with the standard image compactness of the aluminum foil defect test image to obtain a compactness difference ratio; The obtained length-width difference ratio and compactness difference ratio are multiplied to obtain the defect deformation value; The defect deformation value is compared with the defect deformation threshold. If the defect deformation value is greater than the defect deformation threshold, it means that the defect point in the test image is more deformed than the defect point in the aluminum foil defect test image. The acquired defect point imaging is unqualified and is marked as an unclear defect point.

5. The aluminum foil detection method based on computer vision according to claim 2, characterized in that: The defect recognition ratio is obtained as follows: The total number of defect points identified in the test image and the total number of defect points in the aluminum foil defect test map are obtained; the total number of defect points identified in the test image and the total number of defect points in the aluminum foil defect test map are processed by ratio to obtain a defect recognition ratio, which is marked as SB.

6. The aluminum foil detection method based on computer vision according to claim 1, characterized in that: The method for obtaining the illumination associated value is as follows: Perform data processing on the reflective area and the identified detail area to obtain the detail performance value XS; Perform data processing on the sub-region identification area to obtain the sub-region deviation ratio FG; The sub-region deviation ratio FG and detail performance value XS are processed and the formula is used , and obtain the illumination correlation value XJ; where s1 and s2 are preset proportional coefficients, and both are greater than zero.

7. The method for detecting aluminum foil based on computer vision according to claim 6, characterized in that: The method for obtaining the detail performance value is as follows: Mark the area where the spots or stripes on the surface of the test image are too bright as the reflective area; obtain the area of ​​the reflective area to obtain the reflective area; Compare the reflective area with the total area of ​​the test image to obtain the reflective area ratio; Obtain the number of unclear defect points in the reflective area, and perform ratio processing on it with the total number of unclear points to obtain the reflective defect ratio; The reflective defect ratio and the reflective area ratio are processed as ratios to obtain the reflective correlation ratio; Get the area of ​​the detail region identified in the test image; The number of unclear defect points in the unidentified detail area is obtained, and the ratio is processed with the total number of unclear points to obtain the detail defect ratio; The reflection correlation ratio and detail defect ratio are weightedly summed to obtain the detail performance value, which is marked as XS.

8. The aluminum foil detection method based on computer vision according to claim 6, characterized in that: The sub-region deviation ratio is obtained as follows: Analyze the deviation of sub-region identification area; The sub-region identification area deviation is compared with the sub-region identification area deviation threshold, the number of sub-region identification area deviations greater than the sub-region identification area deviation threshold is obtained, and it is compared with the total number of sub-region identification area deviations to obtain the sub-region deviation ratio, which is marked as FG.

9. The aluminum foil detection method based on computer vision according to claim 8, characterized in that: The sub-region identification area deviation is obtained in the following manner: The detail area in the test image is evenly divided into several sub-areas, and the area of ​​the detail area identified in the sub-area is obtained, which is marked as the sub-area identification area; Sum and average the identified areas of all sub-regions to obtain the mean of the identified areas of the sub-regions; Subtract the identified areas of all sub-regions from the mean of the identified areas of the sub-regions and take the absolute value to obtain the sub-region identified area deviation.

10. An aluminum foil detection system based on computer vision, comprising the following modules: Test parameter acquisition module: Use a high-resolution industrial camera to collect images of the aluminum foil defect test image; identify the defects of the aluminum foil on the image, analyze the defects on the aluminum foil, and obtain the image similarity value; Test parameter analysis module: compares the image similarity value with the image similarity threshold, and determines whether the test image is qualified based on the comparison result. If it is unqualified, a test failure signal is generated; if it is qualified, a test pass signal is generated; Correlation parameter acquisition module: Based on the test failure signal, the correlation between the test image failure and the device lighting conditions is analyzed to obtain the lighting correlation value; Correlation parameter analysis module: compares the illumination correlation value with the illumination correlation threshold, and determines whether the failure of the test image is related to the illumination condition based on the comparison result. If so, generates an illumination correlation signal and adjusts the illumination condition until the test image passes; Qualified detection module: Based on the test pass signal, the aluminum foil is inspected and the defect points on the aluminum foil are analyzed to obtain the detection defect value; The detected defect value is compared with the detected defect threshold, and whether the produced aluminum foil is qualified is determined based on the comparison result.

Images