Thermal battery electrode plate visual defect detection method and visual defect detection device

Abstract

The invention belongs to the field of machine visual defect detection, and particularly relates to a thermal battery electrode plate visual defect detection method and a visual defect detection device. Product size information of a thermal battery electrode plate is acquired, and the view of a camera is configured; the method comprises the following steps: sucking a to-be-detected thermal battery electrode plate, placing the to-be-detected thermal battery electrode plate on a detection platform, shooting a front image, a side image, a back image and an inner hole image of the thermal battery electrode plate through detection cameras in multiple directions, and processing to obtain to-be-detected images of the front image, the side image, the back image and the inner hole image; and describing and judging defects by adopting the trained YOLO8 recognition model according to the to-be-detected images of the front image, the side image, the back image and the inner hole image.

Description

Technical Field

[0001] The invention belongs to the field of machine vision defect detection, in particular to a thermal battery electrode sheet visual defect detection method and a visual defect detection device. Background Art

[0002] Thermal batteries are also called molten salt batteries and heat-activated storage batteries. When stored, the electrolyte is a non-conductive solid. When used, an electric ignition head or a striker mechanism ignites the heating agent inside, causing the electrolyte in the electrode sheet to melt and become an ion conductor, thus being activated.

[0003] Thermal batteries are a high-temperature energy source with an operating temperature of 350℃~50℃. In addition to the positive and negative poles, the two poles are often separated by an ion-conducting molten electrolyte. Before the electrolyte melts, the battery is inert and can be stored for a long time. Once ignited and activated, it can discharge at high power, with an operating current of several amperes per square centimeter for several seconds, or it can discharge at low power for a long time, lasting 1 hour or more.

[0004] The processing and production of thermal battery electrode sheets require low temperature, fire prevention, and moisture resistance, and the requirements for the process level and environment are very high, and the production is extremely difficult. In the production process. It is necessary to detect the defects of thermal battery electrode sheets, but the existing methods are all through manual visual inspection, which has low detection efficiency, and after the electrode sheets are inspected, it is necessary to manually classify and mark the qualified and unqualified products. The entire inspection process is labor-intensive and has the risk of damaging the electrode sheets. Therefore, the present invention provides a visual defect detection method and device for thermal battery electrode sheets. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for visual defect detection of thermal battery electrode sheets, which has high measurement accuracy and fast response and can be effectively used for defect detection of thermal battery electrode sheets.

[0006] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: a method for visual defect detection of thermal battery electrode sheets, comprising the following steps:

[0007] S1: Obtain product size information of the thermal battery electrode sheet and configure the camera field of view;

[0008] S2: The thermal battery electrode sheet to be inspected is placed on the inspection platform and is placed directly below the front inspection camera. The front inspection camera is used to capture the front image of the thermal battery electrode sheet and process the image to be inspected.

[0009] S3: rotating the thermal battery electrode sheet four times, and sequentially photographing the side images of the thermal battery electrode sheet, and processing them to obtain the image to be detected of the side image;

[0010] S4: The electrode sheet to be inspected is sucked and moved to the top of the back detection camera, and the back image of the thermal battery electrode sheet is captured and processed to obtain the back image to be inspected;

[0011] S5: placing the electrode sheet to be tested in the testing tank, moving the testing tank to the position directly below the inner hole camera, taking an inner hole image, and processing the image to be tested of the inner hole image;

[0012] S6: Based on the images to be inspected, namely, the front image, the side image, the back image and the inner hole image, the defects are described and determined using the trained recognition model.

[0013] Execute step S2 and step S4 to obtain the image to be detected of the front image and the image to be detected of the back image, specifically:

[0014] S2-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI;

[0015] S2-2: Perform Hough circle detection on the binary image to obtain the center and radius information of all circles that meet the conditions;

[0016] S2-3: Cut the region of interest (ROI) image according to the center position and radius information to obtain the image between the inner hole and the outer contour of the thermal battery;

[0017] S2-4: Cut the image between the inner hole and the outer contour of the thermal battery by high threshold and low threshold to obtain the defect area image.

[0018] In step 2-2, the center and radius information of all circles that meet the conditions are obtained, specifically:

[0019] (1) Detect the outer edge of the thermal battery electrode image and find the positions of all circle centers;

[0020] (2) Calculate the distance from other pixels based on the center of the outer edge to obtain the radius;

[0021] (3) Detect the inner holes of the thermal battery electrode image and find the positions of all the circle centers;

[0022] (4) Calculate the distance from the inner hole center to other pixel points to obtain the radius;

[0023] (5) Based on the product size information of the thermal battery electrode sheet obtained in step S1, screen whether there is an inner hole circle and an outer contour circle of the same size.

[0024] The step S2-4 is specifically as follows:

[0025] a. Detect the pixels in the image that are greater than the high threshold and obtain the highlight area;

[0026] b. Detect the pixels in the image that are smaller than the low threshold and obtain the low-brightness area;

[0027] c. Through two corrosion operations and two expansion operations, the interference in the image is filtered out;

[0028] d. Merge the highlight area and the low-brightness area to obtain the image to be detected.

[0029] The step S3 is specifically:

[0030] S3-1: According to the radius information, the ROI position of the thermoelectric battery electrode sheet is determined, that is, the original image is cut with the center of the image as the reference, 2.4 times the radius as the length, and 0.3 of the radius as the width to determine the ROI of the region of interest;

[0031] S3-2: Filter out interference in the image through one erosion operation and one dilation operation;

[0032] S3-3: Cut the highlight area in the image by a high threshold to obtain the image to be detected;

[0033] S3-4: Rotate the electrode sheet and repeatedly perform steps S3-1 to S3-3, that is, take four pictures in total to cover the entire side of the electrode sheet.

[0034] The step S5 is specifically:

[0035] S5-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI of the region of interest;

[0036] S5-2: Filter out interference in the image through one erosion operation and one dilation operation;

[0037] S5-3: Cut out the highlight area in the image through a high threshold to obtain the image to be detected.

[0038] The trained recognition model includes: a detection model and a classification model;

[0039] The trained recognition model makes a determination based on any one of the support vector machine and neural network results.

[0040] The defect description and determination using the trained recognition model is as follows:

[0041] Wherein, the recognition model is Yolov8 deep learning model;

[0042] S6-1: By inputting the images to be inspected into the Yolov8 deep learning model in sequence and then performing forward propagation calculation, the defect information in the image is obtained, namely: the defect size and the confidence of each defect;

[0043] S6-2: Compare the confidence with the manually set fixed confidence threshold, segment the defect information, and if defect information greater than the confidence threshold appears, it is determined that there is a defect; then compare it with the set defect size threshold, if it is greater than the defect size threshold, it is determined that there is an unacceptable defect, and the defect type, defect size and number of defects are recorded; if it is less than the defect size threshold, it is determined to be an acceptable defect, and the qualified information is recorded;

[0044] Defect information that is less than the confidence threshold is judged as an acceptable defect or no defect, and the qualified information is recorded.

[0045] The lighting method of the photographed image is any one of annular dome lighting and side arc lighting.

[0046] A thermal battery electrode sheet visual defect detection device, comprising: a frame, an industrial camera, a camera light source, a light source bracket, a detection tank, a manipulator, a PC, a light source controller, a PLC, a rotating workbench, a loading station and a transfer module;

[0047] The industrial cameras include: a front industrial camera, a side industrial camera, a back industrial camera and an inner hole industrial camera;

[0048] The camera light source includes: a front camera light source, a side camera light source, a back camera light source and an inner hole camera light source;

[0049] The light source bracket includes: a front camera light source bracket, a side camera light source bracket, a back camera light source bracket and an inner hole camera light source bracket;

[0050] The transfer module is arranged on the frame table; the robot is arranged on the transfer module and connected to the PLC to absorb the thermal battery electrode sheet and move it above the frame table;

[0051] The loading station is arranged on the left side of the frame table and is located at the leftmost bottom of the transfer module, and is used for the manipulator to absorb the hot battery electrode sheet transmitted from the upper station;

[0052] The front camera light source bracket is arranged on the frame table and on the right side of the loading station; the front industrial camera is arranged on the top of the front camera light source bracket, and the field of view direction covers the front of the thermal battery electrode sheet; the front camera light source is arranged around the outer circle of the lens of the front industrial camera; the rotating workbench is arranged on the frame table and directly below the front industrial camera;

[0053] The side camera light source bracket is arranged on the frame table and on the right side of the front camera light source bracket; the side industrial camera is arranged on the side camera light source bracket, and the field of view direction covers the side of the thermal battery electrode sheet; the side camera light source is arranged on the right side of the rotating workbench; the side camera light source bracket carries the side industrial camera to move forward and backward to take images of electrode sheets of different models;

[0054] The back camera light source bracket is arranged on the frame table and below the right side of the transfer module; the back camera light source is arranged around the outer circle of the lens of the back industrial camera, and the field of view direction of the back industrial camera covers the back side of the thermal battery electrode sheet;

[0055] The inner hole camera light source bracket is arranged on the frame table and on the right bracket of the transfer module;

[0056] The inner hole industrial camera is arranged on the inner hole camera light source bracket, and the field of view direction covers the inner hole of the electrode sheet; the inner hole camera light source is arranged around the outer circle of the lens of the inner hole industrial camera;

[0057] The detection tank is arranged on the frame table and connected to the PLC, and is used to carry the electrode sheet and move it to the bottom of the inner hole industrial camera. After the detection is completed, the electrode sheet is sucked away by the lower-level workstation;

[0058] The PC is connected to the light source controller, PLC and industrial camera respectively, and is used to process the electrode sheet image sent back by the industrial camera. It also controls the camera light source for lighting through the light source controller and can exchange information with the PLC.

[0059] The present invention has the following beneficial effects and advantages:

[0060] 1. The device of the present invention determines the key feature information of the thermal battery electrode sheet in the image through the recognition model, has fast response and high accuracy, and can be effectively used for defect detection of the thermal battery electrode sheet.

[0061] 2. In the production process of thermal battery electrode sheets, low temperature, fire prevention and moisture resistance are required, so the entire production process is carried out in a low-temperature drying room. This device has application significance in solving the problem of manual inspection difficulties caused by production in a low-temperature drying room.

[0062] 3. Through judgment based on deep learning recognition model, the defect information of thermal batteries can be effectively extracted, which makes up for the lack of accuracy brought by traditional methods, and plays an important role when traditional methods cannot determine defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a structural schematic diagram of an embodiment of a visual defect detection device of the present invention;

[0064] Figure 2This is a schematic diagram of the main detection surface of the thermal battery electrode sheet;

[0065] Figure 3 This is a schematic diagram of the side detection surface of the thermal battery electrode sheet;

[0066] Figure 4 A schematic diagram of a process of visual defect detection method for thermal battery electrode sheets of the present invention;

[0067] Figure 5 A schematic diagram of the structure of a computer device provided by an embodiment of the present invention;

[0068] Among them, 1 is the front industrial camera, 2 is the front camera light source, 3 is the side industrial camera, 4 is the side camera light source, 5 is the back industrial camera, 6 is the back camera light source, 7 is the inner hole industrial camera, 8 is the inner hole camera light source, 9 is the inspection slot, 10 is the manipulator, 11 is the PC, 12 is the light source controller, 13 is the PLC, 14 is the rotating workbench, 15 is the loading station, 16 is the transfer module, 17 is the front camera light source bracket, 18 is the side camera light source bracket, 19 is the back camera light source bracket, and 20 is the inner hole camera light source bracket. DETAILED DESCRIPTION

[0069] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0070] like Figure 4 FIG. 1 is a flow chart of a method for detecting visual defects of a thermal battery electrode sheet according to the present invention. The method for detecting visual defects of a thermal battery electrode sheet according to the present invention comprises the following steps:

[0071] S1: Obtain product size information of the thermal battery electrode sheet and configure the camera field of view;

[0072] S2: Pick up the electrode sheet to be tested and place it directly below the detection camera on the front of the detection platform to take a front image of the electrode sheet;

[0073] S2-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI;

[0074] S2-2: Perform Hough circle detection on the binary image to obtain the center and radius information of all circles that meet the conditions.

[0075] The method of obtaining the center and radius information of all circles that meet the conditions is as follows:

[0076] a. Detect the outer edge of the thermal battery electrode image and find the positions of all circle centers;

[0077] b. Calculate the distance from other pixels based on the center of the outer edge to obtain the radius;

[0078] c. Detect the inner holes of the thermal battery electrode image and find the positions of all the circle centers;

[0079] d. Calculate the distance from the center of the inner hole to other pixels to obtain the radius;

[0080] e. Based on the product size information of the thermal battery electrode sheet obtained in step S1, screen whether there is an inner hole circle and an outer contour circle of the same size.

[0081] S2-3: According to the center position and radius information of the circle, the ROI image is cut to obtain the image between the inner hole and the outer contour of the thermal battery.

[0082] S2-4: Cut the image between the inner hole and the outer contour of the thermal battery by high threshold and low threshold to obtain the defect area image, specifically:

[0083] a. Detect the pixels in the image that are greater than the high threshold and obtain the highlight area;

[0084] b. Detect the pixels in the image that are greater than the low threshold and obtain the low-brightness area;

[0085] c. Through two corrosion operations and two expansion operations, the interference in the image is filtered out;

[0086] d. Merge the highlight area and the low-brightness area to obtain the image to be detected.

[0087] S3: Rotate the thermal battery electrode sheet four times and take side images in sequence;

[0088] S3-1: According to the radius information, the ROI position of the thermoelectric battery electrode sheet is determined, that is, the original image is cut with the center of the image as the reference, 2.4 times the radius as the length, and 0.3 of the radius as the width to determine the ROI of the region of interest;

[0089] S3-2: Filter out interference in the image through one erosion operation and one dilation operation;

[0090] S3-3: Cut the highlight area in the image by a high threshold to obtain the image to be detected;

[0091] S3-4: Rotate the electrode sheet and repeat the operations in S3-1, S3-2, and S3-3. Take four pictures in total to cover the entire side of the electrode sheet.

[0092] S4: Pick up the electrode sheet to be tested and move it to the top of the back detection camera to take an image of the back of the electrode sheet;

[0093] S4-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI of the region of interest;

[0094] S4-2: Perform Hough circle detection on the binary image to obtain the center and radius information of all circles that meet the conditions;

[0095] The method of obtaining the center and radius information of all circles that meet the conditions is as follows:

[0096] a. Detect the outer edge of the thermal battery electrode image and find the positions of all circle centers;

[0097] b. Calculate the distance from other pixels based on the center of the outer edge to obtain the radius;

[0098] c. Detect the inner holes of the thermal battery electrode image and find the positions of all the circle centers;

[0099] d. Calculate the distance from the center of the inner hole to other pixels to obtain the radius;

[0100] e. Based on the product size information of the thermal battery electrode sheet obtained in step S1, screen whether there is an inner hole circle and an outer contour circle of the same size.

[0101] S4-3: Cut the region of interest (ROI) image according to the center position and radius information to obtain the image between the inner hole and the outer contour of the thermal battery;

[0102] S4-4: Cut the image between the inner hole and the outer contour of the thermal battery by high threshold and low threshold to obtain the defect area image, specifically:

[0103] a. Detect the pixels in the image that are greater than the high threshold and obtain the highlight area;

[0104] b. Detect the pixels in the image that are greater than the low threshold and obtain the low-brightness area;

[0105] c. Through two corrosion operations and two expansion operations, the interference in the image is filtered out;

[0106] d. Merge the highlight area and the low-brightness area to obtain the image to be detected.

[0107] S5: placing the electrode sheet to be tested in the test tank, moving the test tank to the bottom of the inner hole camera, and taking an inner hole image;

[0108] S5-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI of the region of interest;

[0109] S5-2: Filter out interference in the image through one erosion operation and one dilation operation;

[0110] S5-3: Cut out the highlight area in the image through a high threshold to obtain the image to be detected.

[0111] S6: Use the trained recognition model to describe and determine the defects. Figure 2 The figure shows the main inspection surface of the thermal battery electrode sheet of the present invention. In this embodiment, the thermal battery electrode sheet is a circular multi-layer sheet, which is a graphite sheet, positive electrode powder, isolation powder, negative electrode powder, and graphite sheet from top to bottom. The defects to be inspected include: dents, cracks, dirt, positive and negative electrode powder interconnection, isolation powder breakage, etc.

[0112] The detection model and classification model are both Yolov8 models. The feature image is input into the Yolov8 model deep learning model, and then the forward propagation calculation is performed to obtain the defect information in the image. The defect degree of the thermal battery electrode sheet is determined by segmenting the defect information with a fixed threshold.

[0113] The above steps are specifically divided into the following steps to determine the degree of defects of the thermal battery electrode sheet, specifically:

[0114] S6-1: By inputting the images to be inspected into the Yolov8 deep learning model in sequence and then performing forward propagation calculation, the defect information in the image is obtained, namely: the defect size and the confidence of each defect;

[0115] S6-2: Compare the confidence with the manually set fixed confidence threshold, segment the defect information, and if defect information greater than the confidence threshold appears, it is determined that there is a defect; then compare it with the set defect size threshold, if it is greater than the defect size threshold, it is determined that there is an unacceptable defect, and the defect type, defect size and number of defects are recorded; if it is less than the defect size threshold, it is determined to be an acceptable defect, and the qualified information is recorded;

[0116] Defect information that is less than the confidence threshold is judged as an acceptable defect or no defect, and the qualified information is recorded.

[0117] like Figure 3 As shown, it is a schematic diagram of the side detection surface of the thermal battery electrode sheet of the present invention. Finally, the defects detected include: the positive and negative electrode powders are connected, the isolation powder is broken, etc.

[0118] like Figure 1The figure is a schematic structural diagram of an embodiment of the visual defect detection device of the present invention. The present invention is a thermal battery electrode sheet visual defect detection device, characterized in that it includes: a frame, an industrial camera, a camera light source, a light source bracket, a detection tank (9), a manipulator (10), a PC (11), a light source controller (12), a PLC (13), a rotating worktable (14), a loading station (15) and a transfer module (16);

[0119] The industrial camera comprises: a front industrial camera (1), a side industrial camera (3), a back industrial camera (5) and an inner hole industrial camera (7);

[0120] The camera light source comprises: a front camera light source (2), a side camera light source (4), a back camera light source (6) and an inner hole camera light source (8);

[0121] The light source bracket comprises: a front camera light source bracket (17), a side camera light source bracket (18), a back camera light source bracket (19) and an inner hole camera light source bracket (20);

[0122] The transfer module (16) is arranged on the frame table; the robot (10) is arranged on the transfer module (16) and is connected to the PLC (13) to absorb the thermal battery electrode sheet and move it above the frame table;

[0123] The loading station (15) is arranged on the left side of the frame table and is located at the leftmost bottom of the transfer module (16), and is used for the manipulator (10) to absorb the hot battery electrode sheet transmitted from the upper station;

[0124] The front camera light source bracket (17) is arranged on the frame table and on the right side of the loading station (15); the front industrial camera (1) is arranged on the top of the front camera light source bracket (17), and the field of view direction covers the front of the thermal battery electrode sheet; the front camera light source (2) is arranged around the outer ring of the lens of the front industrial camera (1); the rotating workbench (14) is arranged on the frame table and directly below the front industrial camera (1);

[0125] The side camera light source bracket (18) is arranged on the frame table and on the right side of the front camera light source bracket (17); the side industrial camera (3) is arranged on the side camera light source bracket (18), and the field of view direction covers the side of the thermal battery electrode sheet; the side camera light source (4) is arranged on the right side of the rotating workbench (14); the side camera light source bracket (18) carries the side industrial camera (3) to move forward and backward to capture images of electrode sheets of different models;

[0126] The back camera light source bracket (19) is arranged on the frame table and below the right side of the transfer module (16); the back camera light source (6) is arranged around the outer circle of the lens of the back industrial camera (5), and the field of view direction of the back industrial camera (5) covers the back side of the thermal battery electrode sheet;

[0127] The inner hole camera light source bracket (20) is arranged on the frame table and on the right bracket of the transfer module (16);

[0128] The inner hole industrial camera (7) is arranged on the inner hole camera light source bracket (20), and the field of view direction covers the inner hole of the electrode sheet; the inner hole camera light source (8) is arranged around the outer circle of the lens of the inner hole industrial camera (7);

[0129] The detection tank (9) is arranged on the frame table and connected to the PLC (13), and is used to carry the electrode sheet and move it to the bottom of the inner hole industrial camera (7). After the detection is completed, the electrode sheet is sucked away by the lower-level workstation;

[0130] The PC (11) is connected to the light source controller (12), the PLC (13) and the industrial camera respectively, and is used to process the electrode sheet image sent back by the industrial camera, control the camera light source for lighting through the light source controller (12), and can exchange information with the PLC (13).

[0131] Embodiment 1:

[0132] In the specific implementation process, the hot battery electrode sheet can be sucked from the material box by the manipulator and placed directly under the industrial camera of the detection equipment to take a front image. After completion, it is rotated and photographed four times for side images. Then the electrode sheet to be tested is sucked and moved to the top of the back detection camera to take the back image of the electrode sheet. Continue to suck the electrode sheet to be tested and place it in the detection slot, move the detection slot to the bottom of the inner hole camera, and take the inner hole image. Finally, the trained recognition model is used to describe and judge the defects.

[0133] In addition, in order to ensure that the defect image is clear and easy to identify, the image quality needs to be guaranteed. In order to obtain high-quality images, for the inspection of thermal battery electrode sheets, annular dome lighting is used for the front, bottom and inner hole, and arc lighting is used for the side.

[0134] like Figure 5 , which is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention, such as a desktop computer, a laptop computer, an embedded motherboard, an Android all-in-one computer, etc. that can execute a program. The computer device of this embodiment at least includes but is not limited to: a processor and a memory connected via a bus. In this embodiment, the processor is used to run the program stored in the memory to implement the thermal battery electrode sheet visual defect detection method in the embodiment.

Claims

1. A method for visual defect detection of thermal battery electrode sheets, characterized in that: The following steps are involved: S1: Obtain product size information of the thermal battery electrode sheet and configure the camera field of view; S2: The thermal battery electrode sheet to be inspected is placed on the inspection platform and is placed directly below the front inspection camera. The front inspection camera is used to capture the front image of the thermal battery electrode sheet and process the image to be inspected. S3: rotating the thermal battery electrode sheet four times, and sequentially photographing the side images of the thermal battery electrode sheet, and processing them to obtain the image to be detected of the side image; S4: The electrode sheet to be inspected is sucked and moved to the top of the back detection camera, and the back image of the thermal battery electrode sheet is captured and processed to obtain the back image to be inspected; S5: placing the electrode sheet to be tested in the testing tank, moving the testing tank to the position directly below the inner hole camera, taking an inner hole image, and processing the image to be tested of the inner hole image; S6: Based on the images to be inspected, namely, the front image, the side image, the back image and the inner hole image, the defects are described and determined using the trained recognition model.

2. A thermal battery electrode sheet visual defect detection method according to claim 1, characterized in that: Execute step S2 and step S4 to obtain the image to be detected of the front image and the image to be detected of the back image, specifically: S2-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI; S2-2: Perform Hough circle detection on the binary image to obtain the center and radius information of all circles that meet the conditions; S2-3: Cut the region of interest (ROI) image according to the center position and radius information to obtain the image between the inner hole and the outer contour of the thermal battery; S2-4: Cut the image between the inner hole and the outer contour of the thermal battery by high threshold and low threshold to obtain the defect area image.

3. A thermal battery electrode sheet visual defect detection method according to claim 2, characterized in that: In step 2-2, the center and radius information of all circles that meet the conditions are obtained, specifically: (1) Detect the outer edge of the thermal battery electrode image and find the positions of all circle centers; (2) Calculate the distance from other pixels based on the center of the outer edge to obtain the radius; (3) Detect the inner holes of the thermal battery electrode image and find the positions of all the circle centers; (4) Calculate the distance from the inner hole center to other pixel points to obtain the radius; (5) Based on the product size information of the thermal battery electrode sheet obtained in step S1, screen whether there is an inner hole circle and an outer contour circle of the same size.

4. A thermal battery electrode sheet visual defect detection method according to claim 2, characterized in that: The step S2-4 is specifically as follows: a. Detect the pixels in the image that are greater than the high threshold and obtain the highlight area; b. Detect the pixels in the image that are smaller than the low threshold and obtain the low-brightness area; c. Through two corrosion operations and two expansion operations, the interference in the image is filtered out; d. Merge the highlight area and the low-brightness area to obtain the image to be detected.

5. A thermal battery electrode sheet visual defect detection method according to claim 1, characterized in that: The step S3 is specifically: S3-1: According to the radius information, the ROI position of the thermoelectric battery electrode sheet is determined, that is, the original image is cut with the center of the image as the reference, 2.4 times the radius as the length, and 0.3 of the radius as the width to determine the ROI of the region of interest; S3-2: Filter out interference in the image through one erosion operation and one dilation operation; S3-3: Cut the highlight area in the image by a high threshold to obtain the image to be detected; S3-4: Rotate the electrode sheet and repeatedly perform steps S3-1 to S3-3, that is, take four pictures in total to cover the entire side of the electrode sheet.

6. A thermal battery electrode sheet visual defect detection method according to claim 1, characterized in that: The step S5 is specifically: S5-1: Determine the ROI position of the thermoelectric battery electrode sheet according to the radius information, that is, cut the original image with the image center as the radius of 1.2 times the base radius to determine the ROI of the region of interest; S5-2: Filter out interference in the image through one erosion operation and one dilation operation; S5-3: Cut out the highlight area in the image through a high threshold to obtain the image to be detected.

7. A thermal battery electrode sheet visual defect detection method according to claim 1, characterized in that: The trained recognition model includes: a detection model and a classification model; The trained recognition model makes a determination based on any one of the support vector machine and neural network results.

8. A thermal battery electrode sheet visual defect detection method according to claim 6, characterized in that: The defect description and determination using the trained recognition model is as follows: Wherein, the recognition model is Yolov8 deep learning model; S6-1: By inputting the images to be inspected into the Yolov8 deep learning model in sequence and then performing forward propagation calculation, the defect information in the image is obtained, namely: the defect size and the confidence of each defect; S6-2: Compare the confidence with the manually set fixed confidence threshold, segment the defect information, and if defect information greater than the confidence threshold appears, it is determined that there is a defect; then compare it with the set defect size threshold, if it is greater than the defect size threshold, it is determined that there is an unacceptable defect, and the defect type, defect size and number of defects are recorded; if it is less than the defect size threshold, it is determined to be an acceptable defect, and the qualified information is recorded; Defect information that is less than the confidence threshold is judged as an acceptable defect or no defect, and the qualified information is recorded.

9. A thermal battery electrode sheet visual defect detection method according to claim 1, characterized in that: The lighting method of the photographed image is any one of annular dome lighting and side arc lighting.

10. A thermal battery electrode sheet visual defect detection device, characterized in that: include: A frame, an industrial camera, a camera light source, a light source bracket, a detection tank (9), a manipulator (10), a PC (11), a light source controller (12), a PLC (13), a rotary table (14), a loading station (15) and a transfer module (16); The industrial camera comprises: a front industrial camera (1), a side industrial camera (3), a back industrial camera (5) and an inner hole industrial camera (7); The camera light source comprises: a front camera light source (2), a side camera light source (4), a back camera light source (6) and an inner hole camera light source (8); The light source bracket comprises: a front camera light source bracket (17), a side camera light source bracket (18), a back camera light source bracket (19) and an inner hole camera light source bracket (20); The transfer module (16) is arranged on the frame table; the robot (10) is arranged on the transfer module (16) and is connected to the PLC (13) to absorb the thermal battery electrode sheet and move it above the frame table; The loading station (15) is arranged on the left side of the frame table and is located at the leftmost bottom of the transfer module (16), and is used for the manipulator (10) to absorb the hot battery electrode sheet transmitted from the upper station; The front camera light source bracket (17) is arranged on the frame table and on the right side of the loading station (15); the front industrial camera (1) is arranged on the top of the front camera light source bracket (17), and the field of view direction covers the front of the thermal battery electrode sheet; the front camera light source (2) is arranged around the outer ring of the lens of the front industrial camera (1); the rotating workbench (14) is arranged on the frame table and directly below the front industrial camera (1); The side camera light source bracket (18) is arranged on the frame table and on the right side of the front camera light source bracket (17); the side industrial camera (3) is arranged on the side camera light source bracket (18), and the field of view direction covers the side of the thermal battery electrode sheet; the side camera light source (4) is arranged on the right side of the rotating workbench (14); the side camera light source bracket (18) carries the side industrial camera (3) to move forward and backward to capture images of electrode sheets of different models; The back camera light source bracket (19) is arranged on the frame table and below the right side of the transfer module (16); the back camera light source (6) is arranged around the outer circle of the lens of the back industrial camera (5), and the field of view direction of the back industrial camera (5) covers the back side of the thermal battery electrode sheet; The inner hole camera light source bracket (20) is arranged on the frame table and on the right bracket of the transfer module (16); The inner hole industrial camera (7) is arranged on the inner hole camera light source bracket (20), and the field of view direction covers the inner hole of the electrode sheet; the inner hole camera light source (8) is arranged around the outer circle of the lens of the inner hole industrial camera (7); The detection tank (9) is arranged on the frame table and connected to the PLC (13), and is used to carry the electrode sheet and move it to the bottom of the inner hole industrial camera (7). After the detection is completed, the electrode sheet is sucked away by the lower-level workstation; The PC (11) is connected to the light source controller (12), the PLC (13) and the industrial camera respectively, and is used to process the electrode sheet image sent back by the industrial camera, control the camera light source for lighting through the light source controller (12), and can exchange information with the PLC (13).

Images