An automated detection and evaluation method and system for an insulating sheet

Abstract

The present invention relates to the field of image processing technology, and an automatic detection and evaluation method and system for an insulating sheet, including: receiving an insulating sheet detection instruction, starting an image acquisition instrument, setting a thickness range, obtaining an extracted thickness set, an insulating sheet sample set and an insulating sheet sample, collecting a sample image, summarizing the sample images to obtain a sample image set, constructing a gray conversion curve, obtaining the insulating sheet to be detected, collecting a detection image, calculating the maximum detection thickness and the minimum detection thickness, calculating the detection thickness difference, judging the evaluation level, adjusting to obtain an optimized detection model, obtaining an optimized detection frame and a defect area, calculating the defect ratio, when the defect ratio is greater than the ratio threshold, confirming that the evaluation level is the elimination level, and when the defect ratio is less than the ratio threshold, confirming that the evaluation level is the qualified level, thus completing the automatic detection and evaluation of the insulating sheet. The present invention can improve the accuracy of the quality detection of the insulating sheet.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an automatic detection and evaluation method and system for an insulating sheet. Background Art

[0002] With the continuous development of the manufacturing industry, automated testing technology has become an important part of modern industrial production, especially in the quality inspection process of insulating sheets. The automated inspection of insulating sheets plays an important role in production efficiency and product consistency. Through high-precision inspection and evaluation, the quality and accuracy of insulating sheets are ensured. Especially in the mass production and precision processing of insulating sheets, automated inspection plays a decisive role.

[0003] Traditional testing methods for insulation sheets rely on manual inspection or simple mechanized testing. Although these methods can ensure the basic quality of the product, in the case of large-scale production and large quality fluctuations, manual inspection and simple mechanized testing will be limited in efficiency and accuracy, resulting in the inability to detect problems in a timely manner, affecting the overall production quality and testing efficiency. Therefore, how to improve the accuracy of insulation sheet quality testing is an important issue that needs to be solved urgently. Summary of the invention

[0004] The present invention provides an automatic detection and evaluation method and system for an insulating sheet, the main purpose of which is to improve the accuracy of quality detection of the insulating sheet.

[0005] To achieve the above object, the present invention provides an automatic detection and evaluation method for an insulating sheet, comprising:

[0006] receiving an insulation sheet detection instruction, and starting a pre-built image acquisition device based on the insulation sheet detection instruction;

[0007] Set the thickness range, extract the thickness range evenly based on the preset thickness interval, obtain the extracted thickness set, and obtain the insulation sheet sample set according to the extracted thickness set, where the thickness range is , the extracted thicknesses in the extracted thickness set correspond one-to-one to the insulating sheet samples in the insulating sheet sample set;

[0008] Extracting insulating sheet samples from the insulating sheet sample set in sequence, and using the started image acquisition device to acquire images of the extracted insulating sheet samples to obtain sample images, and summarizing the sample images to obtain a sample image set, wherein the image acquisition device is located directly above the insulating sheet samples, and the sample images are RGB images;

[0009] Constructing a grayscale conversion curve based on the sample image set, obtaining an insulating sheet to be detected, and using the image acquisition device to acquire a detection image of the insulating sheet to be detected to obtain a detection image;

[0010] Perform thickness detection on the detected image using the grayscale conversion curve to obtain the detected maximum thickness and the detected minimum thickness;

[0011] Calculate the detected thickness difference based on the detected maximum thickness and the detected minimum thickness, and judge the evaluation level based on the detected thickness difference and the preset uniform thickness threshold. Among them, the evaluation levels include: qualified level and elimination level;

[0012] Compare the detected thickness difference with the uniform thickness threshold;

[0013] If the detected thickness difference is greater than the uniform thickness threshold, confirm that the evaluation level is the elimination level;

[0014] If the detected thickness difference is not greater than the uniform thickness threshold, confirm the optimized detection model;

[0015] Perform target detection on the detected image using the optimized detection model to obtain an optimized detection box, obtain the defect area based on the optimized detection box, calculate the defect ratio based on the defect area, and when the defect ratio is greater than the preset ratio threshold, confirm that the evaluation level is the elimination level;

[0016] When the defect ratio is less than the ratio threshold, confirm that the evaluation level is the qualified level, and complete the automatic detection and evaluation of the insulating sheet.

[0017] Optionally, the constructing the grayscale conversion curve based on the sample image set includes:

[0018] Sequentially extract sample images from the sample image set, and perform grayscale conversion operations on the extracted sample images to obtain sample grayscale images, and collect the sample grayscale images to obtain a sample grayscale image set;

[0019] Sequentially extract sample grayscale images from the sample grayscale image set, and perform grayscale measurement operations on the extracted sample grayscale images based on preset measurement parameters to obtain a plurality of measured grayscale values, calculate the mean value of the plurality of measured grayscale values to obtain a grayscale mean value, and collect the grayscale mean values to obtain a grayscale mean value set, where the grayscale mean values in the grayscale mean value set correspond one-to-one with the extracted thicknesses in the extracted thickness set;

[0020] Construct a grayscale conversion curve according to the grayscale mean value set.

[0021] Optionally, the constructing the grayscale conversion curve according to the grayscale mean value set includes:

[0022] Perform grayscale normalization operations on all the grayscale mean values in the grayscale mean value set to obtain a set of normalized grayscale values, and perform thickness normalization operations on all the extracted thicknesses in the extracted thickness set to obtain a set of normalized thickness values, where the normalized grayscale values in the set of normalized grayscale values correspond one-to-one with the normalized thickness values in the set of normalized thickness values;

[0023] Construct a conversion horizontal axis and a conversion vertical axis based on the normalized gray value and the normalized thickness value, construct a conversion coordinate system based on the conversion horizontal axis and the conversion vertical axis, and construct a gray conversion curve according to the set of normalized gray values, the set of normalized thickness values, and the conversion coordinate system.

[0024] Optionally, the thickness detection of the detection image using the gray conversion curve to obtain the detected maximum thickness and the detected minimum thickness includes:

[0025] Perform a graying operation on the detection image to obtain a detected gray image;

[0026] Based on a preset block area, perform image block division on the detected gray image to obtain a set of block images, calculate the block gray mean value of each block image in the set of block images, and summarize the block gray mean values to obtain a set of block gray mean values;

[0027] Extract the maximum gray mean value and the minimum gray mean value from the set of gray mean values, and perform gray normalization operations on all the block gray mean values in the set of block gray mean values using the maximum gray mean value and the minimum gray mean value to obtain a set of normalized block mean values;

[0028] Perform thickness retrieval on all the normalized block mean values in the set of normalized block mean values based on the gray conversion curve to obtain a set of normalized block thicknesses;

[0029] Obtain the detected maximum thickness and the detected minimum thickness from the set of normalized block thicknesses.

[0030] Optionally, the confirmation of the optimized detection model includes:

[0031] Perform object detection operations on a preset optimized image based on a pre-trained object detection model to obtain an object detection box. Among them, there are cracks and a true detection box in the optimized image, and the object detection box is:

[0032]

[0033] Among them, refers to the object detection box, refers to the predicted center abscissa, refers to the predicted center ordinate, refers to the width of the object detection box, refers to the height of the object detection box;

[0034] The true detection box is:

[0035]

[0036] Among them, refers to the true detection box, refers to the true center abscissa, Refers to the true center ordinate, Refers to the true detection box width, Refers to the true detection box height;

[0037] Calculate the region overlap degree based on the target detection box and the true detection box;

[0038] Obtain the target coordinates and target matrix of the target detection box, and obtain the true coordinates and true matrix of the true detection box;

[0039] Calculate the distribution distance based on the target coordinates, target matrix, true coordinates and true matrix;

[0040] Optimize the target detection model according to the region overlap degree and the distribution distance to obtain an optimized detection model.

[0041] Optionally, the calculating the region overlap degree based on the target detection box and the true detection box includes:

[0042] Calculate the total detection area based on the target detection box and the true detection box, and calculate the total overlapping area based on the target detection box and the true detection box;

[0043] Calculate the region overlap degree based on the total detection area and the total overlapping area:

[0044]

[0045] Wherein, Refers to the region overlap degree, Refers to the total overlapping area, Refers to the total detection area.

[0046] Optionally, the obtaining the target coordinates and target matrix of the target detection box, and obtaining the true coordinates and true matrix of the true detection box includes:

[0047] Obtain the target coordinates based on the target detection box, wherein the target coordinates are:

[0048]

[0049] Wherein, Refers to the target coordinates;

[0050] Obtain the target horizontal expansion value and the target vertical expansion value, and construct a target matrix based on the target horizontal expansion value and the target vertical expansion value:

[0051]

[0052] Wherein, Refers to the target matrix, Refers to the target horizontal expansion value, Refers to a preset adjustment factor. Refers to the target vertical expansion value;

[0053] Obtain the real coordinates based on the real detection box, where the real coordinates are:

[0054]

[0055] Among them, Refers to the real coordinates;

[0056] Obtain the real horizontal expansion value and the real vertical expansion value, and construct a real matrix based on the real horizontal expansion value and the real vertical expansion value:

[0057]

[0058] Among them, Refers to the real matrix, Refers to the real horizontal expansion value, Refers to the real vertical expansion value.

[0059] Optionally, calculate the distribution distance based on the target coordinates, the target matrix, the real coordinates and the real matrix, and the calculation formula is as follows:

[0060]

[0061] Among them, Refers to the distribution distance, Refers to calculating the Euclidean distance, Refers to calculating the trace of the matrix.

[0062] Optionally, optimize the target detection model according to the region coincidence degree and the distribution distance to obtain an optimized detection model, including:

[0063] Construct a comprehensive optimization function based on the region coincidence degree and the distribution distance:

[0064]

[0065] Among them, Refers to the comprehensive optimization function, Refers to a preset balance parameter;

[0066] Calculate the comprehensive optimization value of the target detection box based on the comprehensive optimization function;

[0067] Compare the comprehensive optimization value with a preset comprehensive optimization threshold;

[0068] If it is confirmed that the comprehensive optimization value is not greater than the comprehensive optimization threshold, adjust the target coordinates, the width of the target detection frame, and the height of the target detection frame, and calculate the updated coincidence degree and the updated distribution distance by using the adjusted target coordinates, the adjusted width of the target detection frame, and the adjusted height of the target detection frame, so as to use the updated coincidence degree as the regional coincidence degree and the updated distribution distance as the distribution distance, and return to the above step of constructing the comprehensive optimization function based on the regional coincidence degree and the distribution distance until the comprehensive optimization value is greater than the comprehensive optimization threshold, and obtain the optimized detection model.

[0069] To achieve the above object, the present invention further provides an automatic detection and evaluation system for insulating sheets, including:

[0070] A thickness extraction module, configured to receive an insulating sheet detection instruction and start a pre-built image acquisition device based on the insulating sheet detection instruction;

[0071] Set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtain an insulating sheet sample set according to the extracted thickness set, where the thickness range is and the extracted thickness in the extracted thickness set corresponds to the insulating sheet sample in the insulating sheet sample set one by one;

[0072] A sample image acquisition module, configured to sequentially extract insulating sheet samples from the insulating sheet sample set, and use the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarize the sample images to obtain a sample image set, where the image acquisition device is located directly above the insulating sheet sample, and the sample image is an RGB image;

[0073] A thickness detection module, configured to construct a gray conversion curve based on the sample image set, obtain an insulating sheet to be detected, and use the image acquisition device to perform detection image acquisition on the insulating sheet to be detected to obtain a detection image;

[0074] Use the gray conversion curve to perform thickness detection on the detection image to obtain a detected maximum thickness and a detected minimum thickness;

[0075] A grade evaluation module, configured to calculate a detected thickness difference based on the detected maximum thickness and the detected minimum thickness, and judge the evaluation grade based on the detected thickness difference and a preset uniform thickness threshold, where the evaluation grades include: a qualified grade and a rejection grade;

[0076] Compare the detected thickness difference with the uniform thickness threshold;

[0077] If the detected thickness difference is greater than the uniform thickness threshold, confirm that the evaluation grade is the rejection grade;

[0078] If the detected thickness difference is not greater than the uniform thickness threshold, confirm the optimized detection model;

[0079] Perform object detection on the detection image using an optimized detection model to obtain an optimized detection box, obtain the defect area based on the optimized detection box, calculate the defect ratio based on the defect area, and when the defect ratio is greater than a preset ratio threshold, confirm that the evaluation level is the elimination level;

[0080] When the defect ratio is less than the ratio threshold, confirm that the evaluation level is the qualified level, and complete the automatic detection and evaluation of the insulating sheet.

[0081] To solve the above problems, the present invention also provides an electronic device, which includes:

[0082] A memory that stores at least one instruction;

[0083] A processor that executes the instructions stored in the memory to implement the above-mentioned automatic detection and evaluation method of the insulating sheet.

[0084] To solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one instruction is stored, and the at least one instruction is executed by a processor in an electronic device to implement the above-mentioned automatic detection and evaluation method of the insulating sheet.

[0085] To solve the problems described in the background art, first, a thickness range is set, and the thickness range is evenly extracted based on the thickness interval to obtain an extracted thickness set. By reasonably setting the thickness interval, the sample selection deviation can be reduced. Evenly extracting insulating sheet samples of different thicknesses can cover the entire thickness range, ensuring the comprehensiveness and diversity of the data of the insulating sheet samples, which helps to improve the reliability of the automatic detection of insulating sheets. The extracted thicknesses in the extracted thickness set correspond one by one to the insulating sheet samples in the insulating sheet sample set. Each extracted thickness in the extracted thickness set has a corresponding insulating sheet sample, making the subsequent detection results easy to analyze systematically and ensuring the coherence and consistency of the evaluation. Secondly, a gray conversion curve is constructed based on the sample image set. The gray conversion curve can obtain the thickness of the insulating sheet corresponding to the detection image according to the gray value of the detection image, providing accurate data for the subsequent comparison of the uniform thickness threshold. Then, the thickness of the detection image is detected using the gray conversion curve to obtain the maximum detected thickness and the minimum detected thickness. By detecting the thickness of the detection image based on the relationship between the gray value and the thickness value of the insulating sheet in the gray conversion curve, the maximum and minimum thicknesses of the insulating sheet corresponding to the detection image can be accurately obtained, improving the accuracy of the comparison of the thickness threshold. Then, the evaluation level is determined by the uniform thickness threshold and the detected thickness difference, which can improve the efficiency of the automatic detection of insulating sheets. If the detected thickness difference is not greater than the uniform thickness threshold, the target detection model is optimized to obtain an optimized detection model. By optimizing the target detection model, the accuracy of the target detection model can be improved, and the accuracy of the automatic detection of insulating sheets can be enhanced. Finally, the optimized detection model is used to perform target detection on the detection image to obtain an optimized detection frame. The defect area is obtained based on the optimized detection frame, and the defect ratio is calculated based on the defect area. By performing target detection on the detection image using the optimized detection model, the defect area on the detection image can be accurately obtained. Calculating the defect ratio based on the defect area can ensure that the division of the qualified level and the elimination level is more scientific and reasonable. Therefore, the present invention can improve the accuracy of the quality detection of insulating sheets. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 FIG. is a schematic flow chart of an automatic detection and evaluation method for insulating sheets provided by an embodiment of the present invention;

[0087] Figure 2 FIG. is a functional module diagram of an automatic detection and evaluation system for insulating sheets provided by an embodiment of the present invention;

[0088] Figure 3 FIG. is a schematic structural diagram of an electronic device for implementing the automatic detection and evaluation method for insulating sheets provided by an embodiment of the present invention.

[0089] DESCRIPTION OF THE REFERENCE NUMERALS:

[0090] 1. Electronic device; 10. Processor; 11. Memory; 12. Bus.

[0091] The implementation, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. Detailed implementation manners

[0092] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0093] The embodiments of the present application provide an automated detection and evaluation method for insulating sheets. The execution subject of the automated detection and evaluation method for insulating sheets includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiments of the present application. In other words, the automated detection and evaluation method for insulating sheets can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0094] Refer to Figure 1 As shown, it is a schematic flowchart of the automated detection and evaluation method for insulating sheets provided by an embodiment of the present invention. In this embodiment, the automated detection and evaluation method for insulating sheets includes:

[0095] S1. Receive an insulating sheet detection instruction, start a pre-built image acquisition instrument based on the insulating sheet detection instruction, set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtain an insulating sheet sample set according to the extracted thickness set, where the thickness range is , and the extracted thickness in the extracted thickness set corresponds one-to-one with the insulating sheet sample in the insulating sheet sample set.

[0096] It is understandable that the insulating sheet detection instruction refers to an instruction manually issued for evaluating the insulating sheet. Exemplarily, Xiao Zhang is a staff member of a certain insulating material company. One day, Xiao Zhang needs to evaluate the insulating sheet, so Xiao Zhang issues an insulating sheet detection instruction. The image acquisition instrument refers to an instrument for acquiring images of the insulating sheet. Optionally, the image acquisition instrument is a Basler industrial camera. The thickness range refers to the range of the thickness of the insulating sheet, and the thickness range is , and the thickness interval refers to the interval manually set for obtaining insulating sheets with different thicknesses. Optionally, the thickness interval is , and uniformly extracting the thickness range based on a preset thickness interval means extracting the thickness range using the thickness interval. For example, the thickness interval is , starting from in the thickness range, and every Perform one extraction until the maximum thickness in the thickness range is reached, obtaining , a total of 296 extracted thicknesses, and the maximum thickness is . The extracted thickness set refers to the set composed of the extracted thicknesses. The extracted thickness refers to the thickness of the insulating sheet extracted from the thickness range using the thickness interval, which is the extracted thickness. Obtaining the insulating sheet sample set according to the extracted thickness set means obtaining the insulating sheets corresponding to the extracted thicknesses in the extracted thickness set, and then summarizing the obtained insulating sheets to obtain the insulating sheet sample set. The insulating sheet sample set refers to the set composed of the insulating sheet samples. The insulating sheet sample refers to the insulating sheet with the corresponding thickness obtained according to the extracted thicknesses in the extracted thickness set. For example, is the extracted thickness in the extracted thickness set, and the insulating sheet sample is the insulating sheet corresponding to the extracted thickness. The extracted thickness is , then the insulating sheet sample is the insulating sheet with this thickness.

[0097] S2. Sequentially extract insulating sheet samples from the insulating sheet sample set, and use the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarize the sample images to obtain a sample image set. Among them, the image acquisition device is located directly above the insulating sheet sample, and the sample image is an RGB image.

[0098] It is understandable that the sample image refers to the image obtained after using the image acquisition device to perform image acquisition on the insulating sheet sample. The sample image set refers to the set composed of the sample images. The RGB image refers to a color image represented by a combination of three primary colors: red (R), green (G), and blue (B). The RGB image is a prior art and will not be elaborated here.

[0099] S3. Construct a gray conversion curve based on the sample image set, obtain the insulating sheet to be detected, use the image acquisition device to perform detection image acquisition on the insulating sheet to be detected to obtain a detection image, and use the gray conversion curve to perform thickness detection on the detection image to obtain the detection maximum thickness and the detection minimum thickness.

[0100] It is understandable that the gray conversion curve refers to a curve for obtaining the thickness of the insulating sheet based on the gray value. The insulating sheet to be detected refers to the insulating sheet that needs to be automatically detected. The detection image refers to the image obtained after using the image acquisition device to perform image acquisition on the insulating sheet to be detected. The detection maximum thickness refers to the maximum thickness of the insulating sheet to be detected obtained using the gray conversion curve. The detection minimum thickness refers to the minimum thickness of the insulating sheet to be detected obtained using the gray conversion curve.

[0101] Specifically, constructing the gray conversion curve based on the sample image set includes:

[0102] Extract sample images from the sample image set in sequence, perform grayscale conversion operations on the extracted sample images to obtain sample grayscale images, and collect the sample grayscale images to obtain a sample grayscale image set;

[0103] Extract sample grayscale images from the sample grayscale image set in sequence, perform grayscale measurement operations on the extracted sample grayscale images based on preset measurement parameters to obtain multiple measured grayscale values, calculate the mean value of the multiple measured grayscale values to obtain a grayscale mean value, and collect the grayscale mean values to obtain a grayscale mean value set, where the grayscale mean values in the grayscale mean value set correspond one-to-one with the extraction thicknesses in the extraction thickness set;

[0104] Construct a grayscale conversion curve according to the grayscale mean value set.

[0105] It can be explained that performing a grayscale conversion operation on the extracted sample image means converting the extracted sample image into a grayscale image. Specifically, the conversion method of converting the sample image into a grayscale image is to use Python and the image processing library OpenCV to convert the sample image into a grayscale image, which is prior art and will not be elaborated here. A sample grayscale image refers to the grayscale image obtained after performing grayscale conversion on the extracted sample image. A sample grayscale image set refers to the set composed of sample grayscale images. A measurement parameter refers to a parameter set by humans. The number of times of performing a grayscale measurement operation on the sample grayscale image is equal to the value of the measurement parameter. Optionally, if the measurement parameter is 10, then the number of times of performing a grayscale measurement operation on the sample grayscale image is also 10. Performing a grayscale measurement operation on the extracted sample grayscale image means measuring the grayscale value of the sample grayscale image. Specifically, the operation of measuring the grayscale value is to use Python and the image processing library OpenCV to extract the grayscale value of the sample grayscale image, which is prior art and will not be elaborated here. Multiple measured grayscale values refer to the overall grayscale values of the sample grayscale image obtained after performing grayscale measurement on the sample grayscale image based on the measurement parameter. For example, if there are three pixel points in the sample grayscale image, and the grayscale values of the three pixel points are 10, 11, and 10 respectively, calculate the mean value of the grayscale values of these three pixel points to obtain the grayscale value mean, and use this grayscale value mean as the overall grayscale value of the sample grayscale image. The number of multiple measured grayscale values depends on the measurement parameter, and the value of the measurement parameter is consistent with the number of multiple measured grayscale values. For example, if the measurement parameter is 10, perform 10 grayscale measurement operations on the sample grayscale image to obtain 10 grayscale values. The grayscale mean value refers to the mean value of multiple measured grayscale values, and the grayscale mean value set refers to the set composed of grayscale mean values. The fact that the grayscale mean values in the grayscale mean value set correspond one-to-one with the extraction thicknesses in the extraction thickness set means that the grayscale mean values in the grayscale mean value set correspond to the extraction thicknesses in the extraction thickness set. For example, the extraction thickness is , an insulating sheet sample is obtained based on the extracted thickness, an image of the insulating sheet sample is collected to obtain a sample image, a grayscale operation is performed on the sample image to obtain a sample grayscale image, the measurement parameter is 10, 10 grayscale measurement operations are performed on the sample grayscale image to obtain 10 grayscale values, the average value of these 10 grayscale values is calculated to obtain a grayscale average value, and this grayscale average value corresponds to the extracted thickness in the extracted thickness set.

[0106] Specifically, constructing the grayscale conversion curve according to the grayscale average value set includes:

[0107] Performing a grayscale normalization operation on all grayscale average values in the grayscale average value set to obtain a set of normalized grayscale values, and performing a thickness normalization operation on all extracted thicknesses in the extracted thickness set to obtain a set of normalized thickness values. Among them, the normalized grayscale values in the set of normalized grayscale values correspond one-to-one with the normalized thickness values in the set of normalized thickness values;

[0108] Construct a conversion horizontal axis and a conversion vertical axis based on the normalized grayscale value and the normalized thickness value, construct a conversion coordinate system based on the conversion horizontal axis and the conversion vertical axis, and construct a grayscale conversion curve according to the set of normalized grayscale values, the set of normalized thickness values, and the conversion coordinate system.

[0109] It can be explained that performing a grayscale normalization operation on all grayscale average values in the grayscale average value set means performing a grayscale normalization operation on all grayscale average values in the grayscale average value set using the grayscale normalization formula. The grayscale normalization formula is as follows:

[0110]

[0111] Among them, refers to the normalized grayscale value, refers to the grayscale average value currently undergoing grayscale normalization, refers to the smallest grayscale average value in the grayscale average value set, refers to the largest grayscale average value in the grayscale average value set.

[0112] It can be understood that the normalized grayscale value refers to the grayscale value obtained after performing grayscale normalization on the grayscale average values in the grayscale average value set. The set of normalized grayscale values refers to the set composed of the normalized grayscale values. Performing thickness normalization on all extracted thicknesses in the extracted thickness set means performing thickness normalization on all extracted thicknesses in the extracted thickness set using the thickness normalization formula. The thickness normalization formula is as follows:

[0113]

[0114] Among them, refers to the normalized thickness value, refers to the extracted thickness currently undergoing thickness normalization, It refers to the extraction thickness with the smallest concentration of extraction thicknesses. It refers to the extraction thickness with the largest concentration of extraction thicknesses.

[0115] It can be understood that the normalized thickness value refers to the thickness of the insulating sheet obtained after performing thickness normalization on the extraction thicknesses in the extraction thickness set. The normalized thickness value set refers to the set composed of the normalized thickness values. The conversion horizontal axis is the horizontal axis of the conversion coordinate system, and the conversion horizontal axis is used to represent the normalized gray value. The conversion vertical axis is the vertical axis of the conversion coordinate system, and the conversion vertical axis is used to represent the normalized thickness value. Constructing the gray conversion curve according to the normalized gray value set, the normalized thickness value set, and the conversion coordinate system means using a linear regression model to fit the relationship between the normalized gray value and the normalized thickness value, and then drawing the gray conversion curve according to the relationship between the normalized gray value and the normalized thickness value. Using a linear regression model to fit the relationship between the normalized gray value and the normalized thickness value means using Python to call the linear regression model to fit the relationship between the normalized gray value and the normalized thickness value, obtaining the slope and intercept after fitting, which is the prior art and will not be elaborated here. The slope is used to represent the relationship between the normalized gray value and the normalized thickness value. When the slope is negative, it means that as the normalized gray value increases, the normalized thickness value will decrease. When the slope is positive, it means that as the normalized gray value increases, the normalized thickness value will increase. The intercept refers to the normalized thickness value when the normalized gray value is 0. Drawing the gray conversion curve according to the relationship between the normalized gray value and the normalized thickness value means constructing a linear equation according to the slope and intercept after fitting, and then drawing the gray conversion curve according to the linear equation and the conversion coordinate system. The linear equation is as follows:

[0116]

[0117] Among them, refers to the normalized thickness parameter, refers to the slope, refers to the normalized gray parameter, refers to the intercept.

[0118] It can be explained that the normalized thickness parameter refers to the parameter used to output the normalized thickness value, and the normalized gray parameter refers to the parameter used to input the normalized gray value. For example, when the normalized gray value is 0.2, the normalized gray parameter is used to represent 0.2, and the normalized thickness parameter is used to represent the output of the normalized thickness value obtained after inputting 0.2 into the linear equation.

[0119] Specifically, using the gray conversion curve to perform thickness detection on the detection image to obtain the detected maximum thickness and the detected minimum thickness includes:

[0120] Performing a graying operation on the detection image to obtain a detected gray image;

[0121] Perform image segmentation on the detected grayscale image based on a preset segmentation area to obtain a set of segmented images, calculate the average grayscale value of each segmented image in the set of segmented images, summarize the average grayscale values, and obtain a set of average grayscale values;

[0122] Extract the maximum average grayscale value and the minimum average grayscale value from the set of average grayscale values, and perform grayscale normalization operations on all the average grayscale values in the set of average grayscale values using the maximum average grayscale value and the minimum average grayscale value to obtain a set of normalized average values;

[0123] Perform thickness retrieval on all the normalized average values in the set of normalized average values based on the grayscale conversion curve to obtain a set of normalized thicknesses;

[0124] Obtain the detected maximum thickness and the detected minimum thickness from the set of normalized thicknesses.

[0125] Interpretably, performing the grayscale operation on the detected image is the same as performing the grayscale operation on the extracted sample image. The detected grayscale image refers to the grayscale image obtained after grayscaling the detected image. The segmentation area refers to the area artificially set for performing the segmentation operation on the detected grayscale image. Optionally, the segmentation area is , the set of segmented images refers to the set composed of segmented images. The segmented image refers to the image obtained after segmenting the detected grayscale image using the segmentation area. For example, the segmentation area is , the area of the detected grayscale image is , then the detected grayscale image is divided into 5 images, and these 5 images are the segmented images. The average grayscale value of the segmented image refers to the average value of the grayscale values of the segmented image. The set of average grayscale values of the segmented image refers to the set composed of the average grayscale values of the segmented image. The maximum average grayscale value refers to the largest average grayscale value in the set of average grayscale values. The minimum average grayscale value refers to the smallest average grayscale value in the set of average grayscale values. Performing grayscale normalization operations on all the average grayscale values in the set of average grayscale values means performing grayscale normalization operations on all the average grayscale values in the set of average grayscale values using the above grayscale normalization formula, and in the grayscale normalization formula is the minimum average grayscale value, and in the grayscale normalization formula is the maximum average grayscale value. The set of normalized average values refers to the set composed of the normalized average values. The normalized average value refers to the normalized grayscale obtained after performing grayscale normalization on the average grayscale value of the segmented image. Performing thickness retrieval on all the normalized average values in the set of normalized average values based on the grayscale conversion curve means using the normalized average value to retrieve the corresponding value on the conversion horizontal axis in the grayscale conversion curve, and then matching the corresponding value on the conversion vertical axis according to the retrieved corresponding value on the conversion horizontal axis, and confirming the corresponding value on the conversion vertical axis as the normalized thickness. The set of normalized thicknesses refers to the set composed of the normalized thicknesses.

[0126] S4. Calculate the detected thickness difference based on the detected maximum thickness and the detected minimum thickness, and determine the evaluation level based on the detected thickness difference and a preset uniform thickness threshold. The evaluation levels include: a qualified level and a rejected level. Compare the detected thickness difference with the uniform thickness threshold. If the detected thickness difference is greater than the uniform thickness threshold, confirm that the evaluation level is the rejected level.

[0127] Explainable, the evaluation level is used to evaluate the insulating sheet to be detected. The evaluation levels include: a qualified level and a rejected level. The qualified level means that the detected thickness difference of the insulating sheet to be detected is less than the uniform thickness threshold, and the defect ratio of the detected insulating sheet is less than the ratio threshold. The qualified level is used to indicate that the insulating sheet to be detected meets the usage requirements. The rejected level means that the detected thickness difference of the insulating sheet to be detected is greater than the uniform thickness threshold or the defect ratio of the detected insulating sheet is greater than the ratio threshold. The rejected level is used to indicate that the insulating sheet to be detected does not meet the usage requirements. When the detected thickness difference is not greater than the uniform thickness threshold and the defect ratio is less than the ratio threshold, confirm that the evaluation level of the insulating sheet to be detected is the qualified level. When the detected thickness difference is greater than the uniform thickness threshold or the detected thickness difference is not greater than the uniform thickness threshold but the defect ratio is greater than the ratio threshold, confirm that the evaluation level of the insulating sheet to be detected is the rejected level. The detected thickness difference refers to the difference between the detected maximum thickness and the detected minimum thickness. The uniform thickness threshold is a threshold set manually for judging the detected thickness difference.

[0128] S5. If the detected thickness difference is not greater than the uniform thickness threshold, confirm the optimized detection model.

[0129] Explainable, the target detection model refers to a model that performs defect detection on the detected image and generates a detection box, which is an existing technology and will not be elaborated here. Optionally, the target detection model is the YOLOv5 model. The detection box refers to the rectangular box output by the target detection model after performing defect detection on the detected image. The detection box is used to completely represent the crack in the detected image.

[0130] Specifically, the confirmation of the optimized detection model includes:

[0131] Perform target detection operations on a preset optimized image based on a pre-trained target detection model to obtain a target detection box. Among them, there are cracks and real detection boxes in the optimized image. The target detection box is:

[0132]

[0133] Among them, refers to the target detection box, refers to the predicted center abscissa, refers to the predicted center ordinate, refers to the width of the target detection box, refers to the height of the target detection box;

[0134] The true detection box is:

[0135]

[0136] Among them, refers to the true detection box, refers to the abscissa of the true center, refers to the ordinate of the true center, refers to the width of the true detection box, refers to the height of the true detection box;

[0137] Calculate the regional overlap degree based on the target detection box and the true detection box;

[0138] Obtain the target coordinates and target matrix of the target detection box, and obtain the true coordinates and true matrix of the true detection box;

[0139] Calculate the distribution distance based on the target coordinates, target matrix, true coordinates and true matrix;

[0140] Optimize the target detection model according to the regional overlap degree and the distribution distance to obtain an optimized detection model.

[0141] Interpretably, the optimized image refers to the image of the insulating sheet set artificially, which is used to optimize the target detection model. There are cracks and true detection boxes in the optimized image. The crack refers to the crack existing on the insulating sheet corresponding to the optimized image. The true detection box refers to the detection box generated after artificially detecting the optimized image. The true detection box completely surrounds the crack in the optimized image and is used to optimize the target detection model. The target detection box refers to the detection box generated after the target detection model detects the optimized image. The predicted abscissa of the center refers to the abscissa of the geometric center of the target detection box in the optimized coordinate system. The predicted ordinate of the center refers to the ordinate of the geometric center of the target detection box in the optimized coordinate system. The optimized coordinate system refers to the coordinate system in the optimized image, with the vertex at the upper left corner of the optimized image as the origin. The positive direction of the vertical axis of the optimized coordinate system points directly below the origin of the optimized coordinate system, and the positive direction of the horizontal axis of the optimized coordinate system points directly to the right of the origin of the optimized coordinate system. The width of the target detection box refers to the length from the leftmost side to the rightmost side of the target detection box. The height of the target detection box refers to the length from the uppermost side to the lowermost side of the target detection box. The true abscissa of the center refers to the abscissa of the geometric center of the true detection box in the optimized coordinate system. The true ordinate of the center refers to the ordinate of the geometric center of the true detection box in the optimized coordinate system. The width of the true detection box refers to the length from the leftmost side to the rightmost side of the true detection box. The height of the true detection box refers to the length from the uppermost side to the lowermost side of the true detection box.

[0142] Specifically, the calculating the regional overlap degree based on the target detection box and the true detection box includes:

[0143] Calculate the total detection area based on the target detection box and the ground truth detection box, and calculate the total overlapping area based on the target detection box and the ground truth detection box;

[0144] Calculate the area overlap ratio based on the total detection area and the total overlapping area:

[0145]

[0146] where, denotes the area overlap ratio, denotes the total overlapping area, denotes the total detection area.

[0147] Interpretably, the total detection area refers to the area of the region covered by the target detection box and the ground truth detection box, which is obtained by subtracting the total overlapping area of the target detection box and the ground truth detection box from the sum of the areas of the target detection box and the ground truth detection box. The total overlapping area refers to the area of the intersection of the target detection box and the ground truth detection box. The area overlap ratio refers to the ratio of the total overlapping area to the total detection area, and is used to measure the degree of intersection between the target detection box and the ground truth detection box. The larger the area overlap ratio, the greater the degree of intersection between the target detection box and the ground truth detection box, and the closer the target detection box is to the ground truth detection box.

[0148] Specifically, the obtaining of the target coordinates and the target matrix of the target detection box, and the obtaining of the ground truth coordinates and the ground truth matrix of the ground truth detection box include:

[0149] Obtain the target coordinates based on the target detection box, where the target coordinates are:

[0150]

[0151] where, denotes the target coordinates;

[0152] Obtain the target horizontal expansion value and the target vertical expansion value, and construct the target matrix based on the target horizontal expansion value and the target vertical expansion value:

[0153]

[0154] where, denotes the target matrix, denotes the target horizontal expansion value, denotes a preset adjustment factor, denotes the target vertical expansion value;

[0155] Obtain the ground truth coordinates based on the ground truth detection box, where the ground truth coordinates are:

[0156]

[0157] Among them, refers to the true coordinates;

[0158] Obtain the true horizontal expansion value and the true vertical expansion value, and construct a true matrix based on the true horizontal expansion value and the true vertical expansion value:

[0159]

[0160] Among them, refers to the true matrix, refers to the true horizontal expansion value, refers to the true vertical expansion value.

[0161] Interpretably, the target coordinates refer to the coordinates of the geometric center of the target detection frame. The target horizontal expansion value refers to the variance of the width of the target detection frame in the horizontal direction, which describes the variation range of the width of the target detection frame in the horizontal direction under different environmental conditions. The larger the target horizontal expansion value, the larger the variation range of the width of the target detection frame in the horizontal direction. The environmental conditions refer to the conditions of the surrounding environment during the automatic detection of the insulating sheet. Optionally, the environmental condition is the temperature of the environment. The target vertical expansion value refers to the variance of the height of the target detection frame in the vertical direction, which describes the variation range of the height of the target detection frame in the vertical direction under different environmental conditions. The larger the target vertical expansion value, the larger the variation range of the height of the target detection frame in the vertical direction. For example, the measured values of the width of the target detection frame under different environmental conditions are: 98, 100, 102, 101, and 99, with the unit of pixel. Based on this set of measured values, the variance is calculated to be 2.4, so the target horizontal expansion value is 2.4. The calculation method of the target vertical expansion value is the same as that of the target horizontal expansion value, which will not be elaborated here. The adjustment factor refers to a parameter artificially set to adjust the relationship between the width and height of the detection frame. The range of the adjustment factor is from -1 to 1. Optionally, when the adjustment factor is 0, it means that there is no linear relationship between the width and height of the detection frame, that is, the width and height of the detection frame are independent. When the adjustment factor is 1, it means that the width and height of the detection frame will change proportionally, and the detection frame will maintain a fixed aspect ratio. The aspect ratio refers to the ratio of the height to the width of the detection frame. The target matrix refers to the matrix representing the target horizontal expansion value and the target vertical expansion value. For example, if the target horizontal expansion value is 150, the target vertical expansion value is 50, and the adjustment factor is 0.8, then the target matrix is:

[0162] It is understandable that the true coordinates refer to the coordinates of the geometric center of the true detection box, and the true horizontal expansion value refers to the variance of the width of the true detection box in the horizontal direction, which describes the range of variation of the width of the true detection box in the horizontal direction under different environmental conditions. The larger the true horizontal expansion value, the larger the range of variation of the width of the true detection box in the horizontal direction. Moreover, the acquisition method of the true horizontal expansion value is the same as that of the target horizontal expansion value, which will not be elaborated here. The true vertical expansion value refers to the variance of the height of the true detection box in the vertical direction, which describes the range of variation of the height of the true detection box in the vertical direction under different environmental conditions. The larger the true vertical expansion value, the larger the range of variation of the height of the true detection box in the vertical direction. Also, the acquisition method of the true vertical expansion value is the same as that of the target vertical expansion value, which will not be elaborated here. The true matrix refers to the matrix representing the true horizontal expansion value and the true vertical expansion value.

[0163] Specifically, calculating the distribution distance based on the target coordinates, target matrix, true coordinates, and true matrix, the calculation formula is as follows:

[0164]

[0165] Among them, refers to the distribution distance, refers to calculating the Euclidean distance, refers to calculating the trace of the matrix.

[0166] It is interpretable that the distribution distance refers to a parameter for measuring the difference between the target detection box and the true detection box. The smaller the distribution distance, the smaller the difference between the target detection box and the true detection box, and the more similar the target detection box and the true detection box are.

[0167] Specifically, optimizing the target detection model according to the region overlap degree and the distribution distance to obtain an optimized detection model includes:

[0168] Constructing a comprehensive optimization function based on the region overlap degree and the distribution distance:

[0169]

[0170] Among them, refers to the comprehensive optimization function, refers to a preset balance parameter;

[0171] Calculating the comprehensive optimization value of the target detection box based on the comprehensive optimization function;

[0172] Comparing the comprehensive optimization value with a preset comprehensive optimization threshold;

[0173] If it is confirmed that the comprehensive optimization value is not greater than the comprehensive optimization threshold, the target coordinates, the width of the target detection box, and the height of the target detection box are adjusted, and the updated overlap degree and the updated distribution distance are calculated using the adjusted target coordinates, the adjusted width of the target detection box, and the adjusted height of the target detection box, so that the updated overlap degree is the regional overlap degree and the updated distribution distance is the distribution distance, and the step of constructing the comprehensive optimization function based on the regional overlap degree and the distribution distance is returned until the comprehensive optimization value is greater than the comprehensive optimization threshold, and an optimized detection model is obtained.

[0174] Interpretably, the comprehensive optimization function refers to a function that combines the regional overlap degree and the distribution distance, which can more comprehensively evaluate the difference between the target detection box and the true detection box. The comprehensive optimization value refers to the output of the comprehensive optimization function. The larger the comprehensive optimization value, the smaller the difference between the target detection box and the true detection box, and the more similar the target detection box is to the true detection box. The balance parameter refers to a parameter that adjusts the influence degree of the regional overlap degree on the comprehensive optimization function. The larger the balance parameter, the greater the influence degree of the regional overlap degree on the comprehensive optimization function. The comprehensive optimization threshold refers to a threshold set artificially for evaluating the target detection box. When the comprehensive optimization value is greater than the comprehensive optimization threshold, it is confirmed that the target detection box is consistent with the true detection box. Adjusting the target coordinates means adjusting the geometric center of the target detection box. The methods for adjusting the target coordinates, the width of the target detection box, and the height of the target detection box are as follows: First, a target coordinate loss function, a detection box width loss function, and a detection box height loss function are constructed. Then, the loss of the target coordinate loss function is minimized, the loss of the detection box width loss function is minimized, and the loss of the detection box height loss function is minimized. Then, the adjustment of the target coordinates, the width of the target detection box, and the height of the target detection box is completed according to the minimized loss of the target coordinate loss function, the minimized loss of the detection box width loss function, and the minimized loss of the detection box height loss function. This is the prior art and will not be elaborated here. The target coordinate loss function refers to a function that uses the Euclidean distance between the target coordinates and the true coordinates as the loss. The detection box width loss function refers to a function that uses the absolute value of the difference between the width of the target detection box and the width of the true detection box as the loss. The detection box height loss function refers to a function that uses the absolute value of the difference between the height of the target detection box and the height of the true detection box as the loss. This is the prior art and will not be elaborated here. For example, the target coordinate loss function is , and the detection box width loss function is: , and the detection box height loss function is: , the comprehensive optimization value is 0.5, and the comprehensive optimization threshold is 0.6. After adjusting the target coordinates, the width of the target detection box, and the height of the target detection box based on the target coordinate loss function, the detection box width loss function, and the detection box height loss function respectively, the obtained comprehensive optimization value is 0.65, which is greater than the comprehensive optimization threshold, and the adjustment of the target coordinates, the width of the target detection box, and the height of the target detection box is completed. The updated overlap degree refers to the area overlap degree calculated using the adjusted target coordinates, the adjusted width of the target detection box, and the adjusted height of the target detection box, and the updated distribution distance refers to the distribution distance calculated using the adjusted target coordinates, the adjusted width of the target detection box, and the adjusted height of the target detection box.

[0175] S6. Use the optimized detection model to perform target detection on the detection image to obtain an optimized detection box, obtain the defect area based on the optimized detection box, calculate the defect ratio based on the defect area. When the defect ratio is greater than the preset ratio threshold, confirm that the evaluation level is the elimination level. When the defect ratio is less than the ratio threshold, confirm that the evaluation level is the qualified level, and complete the automatic detection and evaluation of the insulating sheet.

[0176] Interpretable. The optimized detection box refers to the detection box generated after the optimized detection model performs target detection on the detection image. The defect area refers to the area of the optimized detection box, the defect ratio refers to the ratio of the defect area to the area of the detection image, and the ratio threshold refers to the threshold set by humans for evaluating the defect ratio.

[0177] To solve the problems described in the background art, first, a thickness range is set, and the thickness range is evenly extracted based on the thickness interval to obtain an extracted thickness set. By reasonably setting the thickness interval, the sample selection deviation can be reduced. Evenly extracting insulating sheet samples of different thicknesses can cover the entire thickness range, ensuring the comprehensiveness and diversity of the data of the insulating sheet samples, which helps to improve the reliability of the automatic detection of insulating sheets. The extracted thicknesses in the extracted thickness set correspond one by one to the insulating sheet samples in the insulating sheet sample set. Each extracted thickness in the extracted thickness set has a corresponding insulating sheet sample, making the subsequent detection results easy to analyze systematically and ensuring the coherence and consistency of the evaluation. Secondly, a gray conversion curve is constructed based on the sample image set. The gray conversion curve can obtain the thickness of the insulating sheet corresponding to the detection image according to the gray value of the detection image, providing accurate data for the subsequent comparison of the uniform thickness threshold. Then, the thickness of the detection image is detected using the gray conversion curve to obtain the maximum detected thickness and the minimum detected thickness. By detecting the thickness of the detection image based on the relationship between the gray value and the thickness value of the insulating sheet in the gray conversion curve, the maximum and minimum thicknesses of the insulating sheet corresponding to the detection image can be accurately obtained, improving the accuracy of the comparison of the thickness threshold. Then, the evaluation level is determined by the uniform thickness threshold and the detected thickness difference, which can improve the efficiency of the automatic detection of insulating sheets. If the detected thickness difference is not greater than the uniform thickness threshold, the target detection model is optimized to obtain an optimized detection model. By optimizing the target detection model, the accuracy of the target detection model can be improved, and the accuracy of the automatic detection of insulating sheets can be increased. Finally, the optimized detection model is used to perform target detection on the detection image to obtain an optimized detection frame. The defect area is obtained based on the optimized detection frame, and the defect ratio is calculated based on the defect area. By performing target detection on the detection image using the optimized detection model, the defect area on the detection image can be accurately obtained. Calculating the defect ratio based on the defect area can ensure that the division of the qualified level and the elimination level is more scientific and reasonable. Therefore, the present invention can improve the accuracy of the quality detection of insulating sheets.

[0178] As Figure 2 shown, it is a functional module diagram of an automatic detection and evaluation system for insulating sheets provided by an embodiment of the present invention.

[0179] The automatic detection and evaluation system 100 for insulating sheets according to the present invention can be installed in an electronic device. According to the functions achieved, the automatic detection and evaluation system 100 for insulating sheets can include a thickness extraction module 101, a sample image acquisition module 102, a thickness detection module 103, and a level evaluation module 104. The modules in the present invention can also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0180] The thickness extraction module 101 is configured to receive an insulating sheet detection instruction and start a pre-built image acquisition device based on the insulating sheet detection instruction;

[0181] Set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extraction thickness set, and obtain an insulating sheet sample set according to the extraction thickness set, where the thickness range is , and the extraction thickness in the extraction thickness set corresponds one-to-one with the insulating sheet sample in the insulating sheet sample set;

[0182] The sample image acquisition module 102 is configured to sequentially extract insulating sheet samples from the insulating sheet sample set, and use the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarize the sample images to obtain a sample image set, where the image acquisition device is located directly above the insulating sheet sample, and the sample image is an RGB image;

[0183] The thickness detection module 103 is configured to construct a gray conversion curve based on the sample image set, obtain the insulating sheet to be detected, and use the image acquisition device to perform detection image acquisition on the insulating sheet to be detected to obtain a detection image;

[0184] Use the gray conversion curve to perform thickness detection on the detection image to obtain a detected maximum thickness and a detected minimum thickness;

[0185] The grade evaluation module 104 is configured to calculate a detected thickness difference based on the detected maximum thickness and the detected minimum thickness, and judge the evaluation grade based on the detected thickness difference and a preset uniform thickness threshold, where the evaluation grades include: a qualified grade and a rejected grade;

[0186] Compare the detected thickness difference with the uniform thickness threshold;

[0187] If the detected thickness difference is greater than the uniform thickness threshold, confirm that the evaluation grade is the rejected grade;

[0188] If the detected thickness difference is not greater than the uniform thickness threshold, confirm an optimized detection model;

[0189] Use the optimized detection model to perform object detection on the detection image to obtain an optimized detection frame, obtain a defect area based on the optimized detection frame, calculate a defect ratio based on the defect area, and when the defect ratio is greater than a preset ratio threshold, confirm that the evaluation grade is the rejected grade;

[0190] When the defect ratio is less than the ratio threshold, confirm that the evaluation grade is the qualified grade, and complete the automatic detection and evaluation of the insulating sheet.

[0191] Specifically, each module in the automatic detection and evaluation system 100 of the insulating sheet in the embodiment of the present invention adopts the same as the aboveFigure 1 The same technical means as the automated detection and evaluation method of the insulating sheet described in

[0192] As Figure 3 shown, it is a schematic structural diagram of an electronic device for implementing the automated detection and evaluation method of the insulating sheet provided by an embodiment of the present invention.

[0193] The electronic device 1 may include a processor 10, a memory 11, and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as the automated detection and evaluation method program of the insulating sheet.

[0194] Among them, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (such as: SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, etc. The memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as the mobile hard disk of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the electronic device 1. Further, the memory 11 further includes the internal storage unit of the electronic device 1 and also includes an external storage device. The memory 11 can not only be used to store application software installed in the electronic device 1 and various types of data, such as the code of the automated detection and evaluation method program of the insulating sheet, etc., but also be used to temporarily store data that has been output or will be output.

[0195] The processor 10 may be composed of integrated circuits in some embodiments. For example, it may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits with the same or different functions, including a combination of one or more Central Processing Units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control core (Control Unit) of the electronic device, connecting various components of the entire electronic device through various interfaces and lines, and by running or executing programs or modules stored in the memory 11 (such as the automated detection and evaluation method program of the insulating sheet, etc.), and calling data stored in the memory 11, to perform various functions of the electronic device 1 and process data.

[0196] The bus 12 may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus 12 can be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to enable connection and communication between the memory 11 and at least one processor 10, etc.

[0197] Figure 3 Only an electronic device with components is shown. Those skilled in the art can understand that Figure 3 the shown structure does not constitute a limitation on the electronic device 1, and it may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0198] For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for supplying power to each component. Preferably, the power source can be logically connected to the at least one processor 10 through a power management device, so as to implement functions such as charging management, discharging management, and power consumption management through the power management device. The power source may also include any components such as one or more DC or AC power sources, a recharge device, a power failure detection circuit, a power converter or inverter, a power status indicator, etc. The electronic device 1 may also include various sensors, a Bluetooth module, a Wi-Fi module, etc., which will not be elaborated here.

[0199] Furthermore, the electronic device 1 may further include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0200] Optionally, the electronic device 1 may further include a user interface. The user interface may be a display, an input unit (such as a keyboard), and optionally, the user interface may also be a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch liquid crystal display, and an OLED (Organic Light-Emitting Diode) toucher, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display the information processed in the electronic device 1 and to display a visual user interface.

[0201] The program of the automatic detection and evaluation method of the insulating sheet stored in the memory 11 in the electronic device 1 is a combination of multiple instructions. When running in the processor 10, it can achieve:

[0202] Receive an insulating sheet detection instruction, and start a pre-built image acquisition device based on the insulating sheet detection instruction;

[0203] Set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtain an insulating sheet sample set according to the extracted thickness set. Among them, the thickness range is , and the extracted thickness in the extracted thickness set corresponds one by one to the insulating sheet sample in the insulating sheet sample set;

[0204] Sequentially extract insulating sheet samples from the insulating sheet sample set, and use the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarize the sample images to obtain a sample image set. Among them, the image acquisition device is located directly above the insulating sheet sample, and the sample image is an RGB image;

[0205] Construct a grayscale conversion curve based on the sample image set, obtain the insulating sheet to be detected, and use the image acquisition device to perform detection image acquisition on the insulating sheet to be detected to obtain a detection image;

[0206] Use the grayscale conversion curve to perform thickness detection on the detection image to obtain the maximum detection thickness and the minimum detection thickness;

[0207] Calculate the detection thickness difference based on the maximum detection thickness and the minimum detection thickness, and judge the evaluation level based on the detection thickness difference and a preset uniform thickness threshold. Among them, the evaluation levels include: qualified level and elimination level;

[0208] Compare the detection thickness difference with the uniform thickness threshold;

[0209] If the detection thickness difference is greater than the uniform thickness threshold, confirm that the evaluation level is the elimination level;

[0210] If the detection thickness difference is not greater than the uniform thickness threshold, confirm an optimized detection model;

[0211] Use the optimized detection model to perform object detection on the detection image to obtain an optimized detection frame, obtain the defect area based on the optimized detection frame, calculate the defect ratio based on the defect area, and when the defect ratio is greater than a preset ratio threshold, confirm that the evaluation level is the elimination level;

[0212] When the defect ratio is less than the ratio threshold, confirm that the evaluation level is the qualified level, and complete the automatic detection and evaluation of the insulating sheet.

[0213] Specifically, the specific implementation method of the above instructions by the processor 10 can refer toFigures 1 to 3 Descriptions of relevant steps in corresponding embodiments are not elaborated herein.

[0214] Furthermore, if the modules / units integrated in the electronic device 1 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory).

[0215] The present invention also provides a computer-readable storage medium. The readable storage medium stores a computer program. When the computer program is executed by a processor of an electronic device, it can implement:

[0216] Receiving an insulating sheet detection instruction, and starting a pre-built image acquisition device based on the insulating sheet detection instruction;

[0217] Setting a thickness range, uniformly extracting the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtaining an insulating sheet sample set according to the extracted thickness set, where the thickness range is and the extracted thicknesses in the extracted thickness set correspond one by one to the insulating sheet samples in the insulating sheet sample set;

[0218] Sequentially extracting insulating sheet samples from the insulating sheet sample set, and using the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarizing the sample images to obtain a sample image set, where the image acquisition device is located directly above the insulating sheet sample, and the sample images are RGB images;

[0219] Constructing a gray-scale conversion curve based on the sample image set, obtaining an insulating sheet to be detected, and using the image acquisition device to perform detection image acquisition on the insulating sheet to be detected to obtain a detection image;

[0220] Performing thickness detection on the detection image using the gray-scale conversion curve to obtain a detected maximum thickness and a detected minimum thickness;

[0221] Calculating a detected thickness difference based on the detected maximum thickness and the detected minimum thickness, and judging an evaluation level based on the detected thickness difference and a preset uniform thickness threshold, where the evaluation levels include: a qualified level and a rejection level;

[0222] Comparing the detected thickness difference with the uniform thickness threshold;

[0223] If the detected thickness difference is greater than the uniform thickness threshold, it is confirmed that the evaluation level is the rejection level;

[0224] If the detected thickness difference is not greater than the uniform thickness threshold, an optimized detection model is confirmed;

[0225] Use the optimized detection model to perform object detection on the detected image to obtain an optimized detection box, obtain the defect area based on the optimized detection box, calculate the defect ratio based on the defect area, and when the defect ratio is greater than the preset ratio threshold, confirm that the evaluation level is the elimination level;

[0226] When the defect ratio is less than the ratio threshold, confirm that the evaluation level is the qualified level, and complete the automatic detection and evaluation of the insulating sheet.

[0227] In several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative, and there may be other division methods in actual implementation.

[0228] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0229] In addition, in each embodiment of the present invention, the functional modules can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware, or in the form of a hardware plus software functional module.

[0230] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0231] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automated detection and evaluation method for an insulating sheet, characterized in that: The method comprises: receiving an insulation sheet detection instruction, and starting a pre-built image acquisition device based on the insulation sheet detection instruction; Set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtain an insulating sheet sample set based on the extracted thickness set, wherein the thickness range is [0.05, 3] mm, and the extracted thicknesses in the extracted thickness set correspond one-to-one to the insulating sheet samples in the insulating sheet sample set; Extracting insulating sheet samples from the insulating sheet sample set in sequence, and using the started image acquisition device to acquire images of the extracted insulating sheet samples to obtain sample images, and summarizing the sample images to obtain a sample image set, wherein the image acquisition device is located directly above the insulating sheet samples, and the sample images are RGB images; Constructing a grayscale conversion curve based on the sample image set, obtaining an insulating sheet to be detected, and using the image acquisition device to acquire a detection image of the insulating sheet to be detected to obtain a detection image; The step of constructing a grayscale conversion curve comprises: A grayscale normalization operation is performed on all grayscale means in the grayscale mean value set to obtain a normalized grayscale value set, and a thickness normalization operation is performed on all extracted thicknesses in the extracted thickness set to obtain a normalized thickness value set, wherein the normalized grayscale values ​​in the normalized grayscale value set correspond one to one to the normalized thickness values ​​in the normalized thickness value set; Construct a conversion horizontal axis and a conversion vertical axis according to the normalized gray value and the normalized thickness value, construct a conversion coordinate system based on the conversion horizontal axis and the conversion vertical axis, and construct a gray conversion curve according to the normalized gray value set, the normalized thickness value set and the conversion coordinate system; Performing thickness detection on the detection image using the grayscale conversion curve to obtain a maximum detection thickness and a minimum detection thickness, wherein the maximum detection thickness refers to the maximum thickness of the insulating sheet to be detected obtained using the grayscale conversion curve, and the minimum detection thickness refers to the minimum thickness of the insulating sheet to be detected obtained using the grayscale conversion curve; Calculate the difference in thickness based on the maximum thickness and the minimum thickness, and determine the evaluation level based on the difference in thickness and a preset uniform thickness threshold, wherein the evaluation levels include: qualified level and eliminated level; Compare the detected thickness difference with the uniform thickness threshold; If the detected thickness difference is greater than the uniform thickness threshold, the evaluation level is confirmed to be the elimination level; If the detected thickness difference is not greater than the uniform thickness threshold, the optimized detection model is confirmed; Performing target detection on the detection image by using an optimized detection model to obtain an optimized detection frame, obtaining a defect area based on the optimized detection frame, calculating a defect ratio based on the defect area, and confirming that the evaluation level is an elimination level when the defect ratio is greater than a preset ratio threshold; When the defect ratio is less than the ratio threshold, the evaluation level is confirmed to be a qualified level, and the automated inspection and evaluation of the insulating sheet is completed.

2. The automated detection and evaluation method for insulating sheets according to claim 1, characterized in that: The step of constructing a grayscale conversion curve based on the sample image set includes: Extracting sample images from the sample image set in sequence, performing grayscale operations on the extracted sample images to obtain sample grayscale images, and collecting the sample grayscale images to obtain a sample grayscale image set; Extracting sample grayscale images in the sample grayscale image set in sequence, and performing grayscale measurement operations on the extracted sample grayscale images based on preset measurement parameters to obtain a plurality of measured grayscale values, performing mean calculation on the plurality of measured grayscale values ​​to obtain grayscale means, and collecting the grayscale means to obtain a grayscale mean set, wherein the grayscale means in the grayscale mean set correspond one-to-one to the extracted thicknesses in the extracted thickness set; A grayscale conversion curve is constructed according to the grayscale mean value set.

3. The automated detection and evaluation method for insulating sheets according to claim 2, characterized in that: The method of using the grayscale conversion curve to perform thickness detection on the detection image to obtain the maximum detection thickness and the minimum detection thickness includes: Performing a grayscale operation on the detection image to obtain a detection grayscale image; The detection grayscale image is divided into blocks based on a preset block area to obtain a block image set, the block grayscale mean of each block image in the block image set is calculated, and the block grayscale means are summarized to obtain a block grayscale mean set; Extracting the maximum grayscale mean and the minimum grayscale mean from the grayscale mean set, and using the maximum grayscale mean and the minimum grayscale mean to perform grayscale normalization operation on all block grayscale means in the block grayscale mean set to obtain a normalized block mean set; Based on the grayscale conversion curve, all normalized block means in the normalized block mean set are subjected to thickness retrieval to obtain a normalized block thickness set; The maximum and minimum thicknesses detected are obtained from the normalized block thickness set.

4. The automated detection and evaluation method for insulating sheets according to claim 3, characterized in that: The step of determining the optimized detection model comprises: Based on the pre-trained target detection model, the target detection operation is performed on the preset optimized image to obtain the target detection frame, where there are cracks and real detection frames in the optimized image. The target detection frame is: R1=(a1,s1,d1,f1) Among them, R1 refers to the target detection box, a1 refers to the horizontal coordinate of the prediction center, s1 refers to the vertical coordinate of the prediction center, d1 refers to the width of the target detection box, and f1 refers to the height of the target detection box; The real detection box is: R2=(a2,s2,d2,f2) Among them, R2 refers to the real detection frame, a2 refers to the real center horizontal coordinate, s2 refers to the real center vertical coordinate, d2 refers to the real detection frame width, and f2 refers to the real detection frame height; Calculate the region overlap based on the target detection frame and the real detection frame; Get the target coordinates and target matrix of the target detection frame, and get the real coordinates and real matrix of the real detection frame; Calculate the distribution distance based on the target coordinates, the target matrix, the real coordinates and the real matrix; The target detection model is optimized according to the area overlap and distribution distance to obtain an optimized detection model.

5. The automated detection and evaluation method for insulating sheets according to claim 4, characterized in that: The calculating the region overlap based on the target detection frame and the real detection frame includes: The total detection area is calculated based on the target detection frame and the true detection frame, and the total overlapping area is calculated based on the target detection frame and the true detection frame; Calculate the area overlap based on the total detection area and the total overlapping area: Among them, T refers to the area overlap, T3 refers to the total overlapping area, and T2 refers to the total detection area.

6. The automated detection and evaluation method for insulating sheets according to claim 5, characterized in that: The step of obtaining the target coordinates and the target matrix of the target detection frame and obtaining the real coordinates and the real matrix of the real detection frame includes: Get the target coordinates based on the target detection frame, where the target coordinates are: in, Refers to the target coordinates; Get the target horizontal expansion value and the target vertical expansion value, and build the target matrix based on the target horizontal expansion value and the target vertical expansion value: Among them, ψ1 refers to the target matrix, δ1 2 refers to the target horizontal expansion value, ζ refers to the preset adjustment factor, δ2 2 Refers to the target vertical expansion value; Get the real coordinates based on the real detection frame, where the real coordinates are: in, Refers to the real coordinates; Get the real horizontal expansion value and the real vertical expansion value, and construct the real matrix based on the real horizontal expansion value and the real vertical expansion value: Among them, ψ2 refers to the real matrix, ε1 2 Refers to the true horizontal expansion value, ε2 2 Refers to the true vertical expansion value.

7. The automated detection and evaluation method for insulating sheets according to claim 6, characterized in that: The distribution distance is calculated based on the target coordinates, target matrix, real coordinates and real matrix, and the calculation formula is as follows: Among them, G refers to the distribution distance, ||*|| refers to calculating the Euclidean distance, and λ(*) refers to calculating the trace of the matrix.

8. The automated detection and evaluation method for insulating sheets according to claim 7, characterized in that: The target detection model is optimized according to the area overlap and distribution distance to obtain an optimized detection model, including: A comprehensive optimization function is constructed based on the regional overlap and distribution distance: H=θ×T+(1-θ)×G Among them, H refers to the comprehensive optimization function, and θ refers to the preset balance parameter; Calculating a comprehensive optimization value of the target detection frame based on a comprehensive optimization function; Compare the comprehensive optimization value with the preset comprehensive optimization threshold; If it is confirmed that the comprehensive optimization value is not greater than the comprehensive optimization threshold, the target coordinates, the target detection frame width and the target detection frame height are adjusted, and the updated overlap and the updated distribution distance are calculated using the adjusted target coordinates, the adjusted target detection frame width and the adjusted target detection frame height, with the updated overlap being the regional overlap and the updated distribution distance being the distribution distance, and returning to the above step of constructing a comprehensive optimization function based on the regional overlap and the distribution distance until the comprehensive optimization value is greater than the comprehensive optimization threshold to obtain an optimized detection model.

9. An automated detection and evaluation system for insulating sheets, characterized in that: The system comprises: A thickness extraction module, used for receiving an insulation sheet detection instruction and starting a pre-built image acquisition device based on the insulation sheet detection instruction; Set a thickness range, uniformly extract the thickness range based on a preset thickness interval to obtain an extracted thickness set, and obtain an insulating sheet sample set based on the extracted thickness set, wherein the thickness range is [0.05, 3] mm, and the extracted thicknesses in the extracted thickness set correspond one-to-one to the insulating sheet samples in the insulating sheet sample set; A sample image acquisition module is used to sequentially extract insulating sheet samples from the insulating sheet sample set, and use the started image acquisition device to perform image acquisition on the extracted insulating sheet samples to obtain sample images, and summarize the sample images to obtain a sample image set, wherein the image acquisition device is located directly above the insulating sheet samples, and the sample images are RGB images; The thickness detection module is used to construct a grayscale conversion curve based on the sample image set, obtain the insulating sheet to be detected, and use the image acquisition device to collect the detection image of the insulating sheet to be detected to obtain the detection image. The grayscale conversion curve is constructed, including: A grayscale normalization operation is performed on all grayscale means in the grayscale mean value set to obtain a normalized grayscale value set, and a thickness normalization operation is performed on all extracted thicknesses in the extracted thickness set to obtain a normalized thickness value set, wherein the normalized grayscale values ​​in the normalized grayscale value set correspond one to one to the normalized thickness values ​​in the normalized thickness value set; Construct a conversion horizontal axis and a conversion vertical axis according to the normalized gray value and the normalized thickness value, construct a conversion coordinate system based on the conversion horizontal axis and the conversion vertical axis, and construct a gray conversion curve according to the normalized gray value set, the normalized thickness value set and the conversion coordinate system; Performing thickness detection on the detection image using the grayscale conversion curve to obtain a maximum detection thickness and a minimum detection thickness, wherein the maximum detection thickness refers to the maximum thickness of the insulating sheet to be detected obtained using the grayscale conversion curve, and the minimum detection thickness refers to the minimum thickness of the insulating sheet to be detected obtained using the grayscale conversion curve; A grade evaluation module, used for calculating the difference in detected thickness based on the maximum detected thickness and the minimum detected thickness, and judging the evaluation grade based on the difference in detected thickness and a preset uniform thickness threshold, wherein the evaluation grades include: qualified grade and eliminated grade; Compare the detected thickness difference with the uniform thickness threshold; If the detected thickness difference is greater than the uniform thickness threshold, the evaluation level is confirmed to be the elimination level; If the detected thickness difference is not greater than the uniform thickness threshold, the optimized detection model is confirmed; Performing target detection on the detection image by using an optimized detection model to obtain an optimized detection frame, obtaining a defect area based on the optimized detection frame, calculating a defect ratio based on the defect area, and confirming that the evaluation level is an elimination level when the defect ratio is greater than a preset ratio threshold; When the defect ratio is less than the ratio threshold, the evaluation level is confirmed to be a qualified level, and the automated inspection and evaluation of the insulating sheet is completed.

Images