CN114782429A

CN114782429A - Image-based lithium battery defect detection method, device, equipment and storage medium

Info

Publication number: CN114782429A
Application number: CN202210689202.6A
Authority: CN
Inventors: 王秦英; 李斌; 区志伟; 陈柯宇
Original assignee: Shenzhen Phoenix Technology Co ltd
Current assignee: Shenzhen Phoenix Technology Co ltd
Priority date: 2022-06-17
Filing date: 2022-06-17
Publication date: 2022-07-22
Anticipated expiration: 2042-06-17
Also published as: CN114782429B

Abstract

The invention relates to the field of artificial intelligence, and discloses a lithium battery defect detection method, device, equipment and storage medium based on images, which are used for improving the accuracy of lithium battery defect detection. The method comprises the following steps: inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection to obtain an insulating film gap detection result; inputting the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection to obtain an insulating film abnormality detection result; and generating a lithium battery defect detection result corresponding to the lithium battery to be detected according to the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result.

Description

Image-based lithium battery defect detection method, device, equipment and storage medium

Technical Field

The invention relates to the field of artificial intelligence, in particular to a lithium battery defect detection method, a lithium battery defect detection device, lithium battery defect detection equipment and a storage medium based on images.

Background

With the development of battery technology, lithium batteries are widely applied to production and life, in the production process of lithium batteries, three layers of materials, namely positive plates, insulating films and negative plates, can be laminated together for winding, and the positive and negative electrodes of normal batteries need to appear alternately in a certain number.

The existing scheme usually adopts traditional modes such as manual detection to execute millimeter-level defect detection, so that the speed is low, the time consumption is huge, the fineness and the accuracy are all the more, and the defect detection accuracy of the lithium battery is low.

Disclosure of Invention

The invention provides a lithium battery defect detection method, a lithium battery defect detection device, lithium battery defect detection equipment and a storage medium based on images, which are used for improving the accuracy of lithium battery defect detection.

The invention provides a lithium battery defect detection method based on an image, which comprises the following steps: setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to the preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected; denoising and cutting the first detection image to obtain positive and negative images, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model for positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and small; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: wide and narrow gaps; inputting the insulating film image into a preset insulating film abnormity detection model for carrying out insulating film abnormity detection to obtain an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulating film breakage, insulating film loss, and insulating film wrinkle; and extracting defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values, and generating a lithium battery defect detection result corresponding to the lithium battery to be detected.

Optionally, in a first implementation manner of the first aspect of the present invention, the denoising and cropping the first detection image to obtain positive and negative electrode images, and performing denoising and image segmentation processing on the second detection image to obtain an insulating film gap image and an insulating film image includes: inputting the first detection image into a preset EDSR model for noise removal to obtain a first high-definition detection image; cutting the first high-definition detection image according to a preset image size to obtain positive and negative images; inputting the second detection image into the EDSR model for noise removal to obtain a second high-definition detection image; carrying out region detection on the second high-definition detection image to obtain an insulating film region and an insulating film gap region; and carrying out region segmentation on the second high-definition detection image according to the insulating film region and the insulating film gap region to obtain an insulating film gap image and an insulating film image.

Optionally, in a second implementation manner of the first aspect of the present invention, the positive and negative electrode images are input to a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, so as to obtain a positive and negative electrode deviation detection result, where the positive and negative electrode deviation detection result includes: the deviation is large and small, including: inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model, wherein the positive and negative electrode deviation detection model comprises: a Darknet-53 network, a Batch nonillization layer, a hidden layer, and an output layer; inputting the positive and negative electrode images into the Darknet-53 network for feature extraction to obtain positive and negative electrode feature images; inputting the positive and negative characteristic images into the Batch nonallimation layer for Batch normalization processing to obtain normalized characteristic images; inputting the normalized feature image into the hidden layer for feature superposition to obtain a fine-grained feature image; inputting the fine-grained characteristic diagram into the output layer to predict positive and negative electrode deviation points to obtain a positive and negative electrode pixel point predicted image; generating positive and negative electrode deviation detection results corresponding to the positive and negative electrode images according to the positive and negative electrode pixel point predicted image, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and the deviation is small.

Optionally, in a third implementation manner of the first aspect of the present invention, the positive and negative electrode deviation detection results corresponding to the positive and negative electrode images are generated according to the positive and negative electrode pixel point predicted image, where the positive and negative electrode deviation detection results include: the deviation is large and small, including: extracting a plurality of positive pixel points and a plurality of negative pixel points in the positive and negative pixel point predicted image; and carrying out pixel-level classification comparison on the plurality of positive electrode pixel points and the plurality of negative electrode pixel points to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and the deviation is small.

Optionally, in a fourth implementation manner of the first aspect of the present invention, the inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection to obtain an insulating film gap detection result, where the insulating film gap detection result includes: the gap width and the gap width include: inputting the insulating film gap image into a preset insulating film gap detection model, wherein the insulating film gap detection model comprises: the method comprises the following steps of (1) extracting a target feature, an RPN, an ROI Align layer and an FCN; inputting the insulating film gap image into the target feature extraction network for feature extraction to obtain an insulating film gap feature map; inputting the insulating film gap characteristic diagram into the RPN network, and generating a preselected frame corresponding to the insulating film gap characteristic diagram according to preset anchor frame information through the RPN network; inputting the preselected frame and the insulating film gap characteristic diagram into the ROI Align layer, performing characteristic fusion on the preselected frame and the insulating film gap characteristic diagram through the ROI Align layer, and performing segmentation and endpoint pooling on the preselected frame to obtain an insulating film gap characteristic labeled diagram; inputting the insulation film gap characteristic labeling diagram into the FCN, predicting each pixel point of the insulation film gap characteristic labeling diagram through the FCN, obtaining and outputting an insulation film gap pixel point prediction result; generating an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result, wherein the insulating film gap detection result comprises: the gap is wide and the gap is narrow.

Optionally, in a fifth implementation manner of the first aspect of the present invention, the generating an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result includes: the gap width and the gap width include: measuring the gap distance of the prediction result of the insulating film gap pixel point to obtain the target gap width; comparing the target gap width with a preset gap threshold value to obtain a gap comparison result; generating an insulation film gap detection result according to the insulation film gap pixel point prediction result and the gap comparison result, wherein the insulation film gap detection result comprises: the gap is wide and the gap is narrow.

Optionally, in a sixth implementation manner of the first aspect of the present invention, the inputting the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection to obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulating film breakage, insulating film loss, and insulating film wrinkle, including: inputting the insulating film image into a preset insulating film abnormality detection model, wherein the insulating film abnormality detection model comprises: a convolutional network, a plurality of residual error networks, and an output layer; performing convolution operation on the insulating film image through the convolution network to obtain a high-dimensional insulating film image; inputting the high-dimensional insulating film image into the residual error networks for residual error processing to obtain a characteristic insulating film image; inputting the characteristic insulating film image into the output layer to perform insulating film pixel point detection, generating and outputting an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises: insulating film breakage, insulating film loss, and insulating film wrinkles.

The invention provides a lithium battery defect detection device based on images in a second aspect, which comprises: the system comprises an acquisition module, a detection module and a display module, wherein the acquisition module is used for setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to the preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected; the processing module is used for carrying out noise reduction and cutting on the first detection image to obtain positive and negative electrode images, and carrying out noise reduction and image segmentation processing on the second detection image to obtain an insulating film gap image and an insulating film image; a deviation detection module, configured to input the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, so as to obtain positive and negative electrode deviation detection results, where the positive and negative electrode deviation detection results include: the deviation is large and small; a gap detection module, configured to input the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, so as to obtain an insulating film gap detection result, where the insulating film gap detection result includes: wide and narrow gaps; an insulating film detection module, configured to input the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection, so as to obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulation film breakage, insulation film loss, and insulation film wrinkle; and the generating module is used for extracting the defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating the characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values and generating the lithium battery defect detection result corresponding to the lithium battery to be detected.

Optionally, in a first implementation manner of the second aspect of the present invention, the processing module is specifically configured to: inputting the first detection image into a preset EDSR model for noise removal to obtain a first high-definition detection image; cutting the first high-definition detection image according to a preset image size to obtain positive and negative images; inputting the second detection image into the EDSR model for noise removal to obtain a second high-definition detection image; carrying out region detection on the second high-definition detection image to obtain an insulating film region and an insulating film gap region; and performing region segmentation on the second high-definition detection image according to the insulating film region and the insulating film gap region to obtain an insulating film gap image and an insulating film image.

Optionally, in a second implementation manner of the second aspect of the present invention, the deviation detecting module is specifically configured to: a deviation detection unit, configured to input the positive and negative electrode images into a preset positive and negative electrode deviation detection model, where the positive and negative electrode deviation detection model includes: a Darknet-53 network, a Batch nonillization layer, a hidden layer, and an output layer; inputting the positive and negative electrode images into the Darknet-53 network for feature extraction to obtain positive and negative electrode feature images; inputting the positive and negative characteristic images into the Batch non-normalization layer for Batch normalization processing to obtain normalized characteristic images; inputting the normalized feature image into the hidden layer for feature superposition to obtain a fine-grained feature image; inputting the feature map with the fine granularity into the output layer to predict positive and negative deviation points, so as to obtain a predicted image of the positive and negative pixel points; the generating unit is used for generating positive and negative electrode deviation detection results corresponding to the positive and negative electrode images according to the positive and negative electrode pixel point predicted images, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and the deviation is small.

Optionally, in a third implementation manner of the second aspect of the present invention, the generating unit is specifically configured to: extracting a plurality of positive pixel points and a plurality of negative pixel points in the positive and negative pixel point predicted image; and carrying out pixel-level classification comparison on the plurality of positive electrode pixel points and the plurality of negative electrode pixel points to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation.

Optionally, in a fourth implementation manner of the second aspect of the present invention, the gap detection module is specifically configured to: a gap detection unit for inputting the insulating film gap image into a preset insulating film gap detection model, wherein the insulating film gap detection model includes: the method comprises the following steps of (1) extracting a target feature, an RPN, an ROI Align layer and an FCN; inputting the insulating film gap image into the target feature extraction network for feature extraction to obtain an insulating film gap feature map; inputting the insulating film gap characteristic diagram into the RPN network, and generating a preselected frame corresponding to the insulating film gap characteristic diagram according to preset anchor frame information through the RPN network; inputting the preselected frame and the insulating film gap characteristic diagram into the ROI Align layer, performing characteristic fusion on the preselected frame and the insulating film gap characteristic diagram through the ROI Align layer, and performing segmentation and endpoint pooling on the preselected frame to obtain an insulating film gap characteristic labeling diagram; inputting the insulation film gap characteristic labeling diagram into the FCN, predicting each pixel point of the insulation film gap characteristic labeling diagram through the FCN, obtaining an insulation film gap pixel point prediction result and outputting the insulation film gap pixel point prediction result; a result generating unit, configured to generate an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result, where the insulating film gap detection result includes: the gap is wide and the gap is narrow.

Optionally, in a fifth implementation manner of the second aspect of the present invention, the result generating unit is specifically configured to: measuring the gap distance of the prediction result of the insulating film gap pixel point to obtain the target gap width; comparing the target gap width with a preset gap threshold value to obtain a gap comparison result; generating an insulation film gap detection result according to the insulation film gap pixel point prediction result and the gap comparison result, wherein the insulation film gap detection result comprises: the gap is wide and the gap is narrow.

Optionally, in a sixth implementation manner of the second aspect of the present invention, the insulating film detection module is specifically configured to: inputting the insulating film image into a preset insulating film abnormality detection model, wherein the insulating film abnormality detection model comprises: a convolutional network, a plurality of residual error networks, and an output layer; performing convolution operation on the insulating film image through the convolution network to obtain a high-dimensional insulating film image; inputting the high-dimensional insulating film image into the residual error networks for residual error processing to obtain a characteristic insulating film image; inputting the characteristic insulating film image into the output layer to perform insulating film pixel point detection, generating and outputting an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises: insulating film breakage, insulating film loss, and insulating film wrinkling.

The third aspect of the present invention provides an image-based lithium battery defect detection apparatus, comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor calls the instructions in the memory to enable the image-based lithium battery defect detection device to execute the image-based lithium battery defect detection method.

A fourth aspect of the present invention provides a computer-readable storage medium having stored therein instructions, which, when run on a computer, cause the computer to execute the above-mentioned image-based lithium battery defect detection method.

In the technical scheme provided by the invention, image acquisition parameters of an image acquisition terminal are set, a first detection image of a lithium battery to be detected is acquired based on the image acquisition terminal and preset anode and cathode point location information, and a second detection image of the lithium battery to be detected is acquired according to preset insulation film point location information, wherein the first detection image comprises an anode and a cathode of the lithium battery to be detected, and the second detection image comprises an insulation film of the lithium battery to be detected; denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model for positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and small; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: wide and narrow gaps; inputting the insulating film image into a preset insulating film abnormity detection model for carrying out insulating film abnormity detection to obtain an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulation film breakage, insulation film loss, and insulation film wrinkle; and extracting defect areas in the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result, calculating characteristic values of the defect areas, classifying and storing the defect areas according to the characteristic values, and generating a lithium battery defect detection result corresponding to the lithium battery to be detected. According to the invention, three deep learning models are constructed in advance, and the defects of the insulating film, the insulating film gap and the positive and negative electrodes of the lithium battery are detected in a targeted manner through the three deep learning models, so that the accuracy of the defect detection of the lithium battery is improved.

Drawings

FIG. 1 is a schematic diagram of an embodiment of a method for detecting defects of a lithium battery based on an image according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of another embodiment of a method for detecting defects of a lithium battery based on an image according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of an embodiment of an image-based lithium battery defect detection apparatus according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of another embodiment of an image-based lithium battery defect detection apparatus according to an embodiment of the present invention;

fig. 5 is a schematic diagram of an embodiment of an image-based lithium battery defect detection device in an embodiment of the present invention.

Detailed Description

The embodiment of the invention provides a lithium battery defect detection method, a lithium battery defect detection device, lithium battery defect detection equipment and a storage medium based on an image, which are used for improving the accuracy of lithium battery defect detection. The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

For convenience of understanding, a specific flow of an embodiment of the present invention is described below, and referring to fig. 1, an embodiment of a method for detecting defects of a lithium battery based on an image according to an embodiment of the present invention includes:

101. setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of the lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to the preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected;

it can be understood that the execution subject of the present invention may be an image-based lithium battery defect detection apparatus, and may also be a terminal or a server, which is not limited herein. The embodiment of the present invention is described by taking a server as an execution subject.

It should be noted that, in the production process of the power lithium battery, the three layers of materials of the positive plate, the insulating film and the negative plate are laminated together to be wound, and the positive and negative electrodes of the normal battery need to appear alternately, and the number of the positive and negative electrodes is fixed. The elongated is the cathode and the thick is the anode. If the single pole continuously appears or the number of the single pole does not accord with the number of the single pole, the battery can be regarded as having defects, and the quality is controlled by carrying out automatic deviation rectification adjustment in time. Then collecting a first detection image of the lithium battery to be detected according to preset positive and negative point location information based on the set image collection terminal, wherein, the positive and negative pole point location information is used for indicating the positions of the positive and negative poles of the lithium battery to be detected, the server collects a second detection image of the lithium battery to be detected according to the insulation film point location information, the insulation film point location information is used for indicating the positions of the insulation films of the lithium battery to be detected in the positive and negative electrodes, the first detection image comprises the positive and negative electrodes of the lithium battery to be detected, and the second detection image comprises the insulation films of the lithium battery to be detected.

102. Denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image;

in this embodiment, a cropping scale of a detection image is acquired according to a region feature of the detection image, the detection image is partitioned according to the cropping scale, and an overlapping region between multiple partitioned image blocks is set. In this embodiment, the image is partitioned by considering the image area characteristics, the noise reduction processing is performed on each image block, and after the noise reduction processing is performed, the server performs cutting and image segmentation processing on the first detection image and the second detection image after the noise reduction processing, so as to obtain corresponding positive and negative electrode images, insulating film gap images and insulating film images.

103. Inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model for positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation;

it should be noted that the positive and negative electrode deviation detection model mainly includes: the system comprises a Darknet-53 network, a Batch nonillization layer, a hiding layer and an output layer, wherein specifically, a server inputs positive and negative electrode images into the Darknet-53 network for feature extraction to obtain positive and negative electrode feature images, and inputs the positive and negative electrode feature images into the Batch nonillization layer for Batch normalization processing to obtain normalized feature images; inputting the normalized feature image into a hidden layer to perform feature superposition to obtain a fine-grained feature image; inputting the feature graph with fine granularity into an output layer to predict positive and negative deviation points, and obtaining a predicted image of the positive and negative pixel points; generating positive and negative electrode deviation detection results corresponding to the positive and negative electrode images according to the positive and negative electrode pixel point predicted images, and if the deviation of the positive and negative electrode pixels in the positive and negative electrode pixel point predicted images is larger than a preset deviation error range, determining that the positive and negative electrode deviation detection results are large in deviation; and if the deviation of the positive and negative pixel points in the positive and negative pixel point predicted image is smaller than the preset deviation error range, determining that the positive and negative deviation detection result is small in deviation.

104. Inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: the gap is wide and narrow;

it should be noted that the fully-connected split sub-networks are added to the insulating film detection model on the basis of the fast RCNN, the original classification and regression tasks can be changed into a combination of classification, regression and separation, and the insulating film detection model mainly comprises a target feature extraction network, an RPN network, an ROI Align layer and an FCN network. Specifically, the server inputs the insulating film gap image into the insulating film gap detection. Extracting an insulating film gap detection result into a target characteristic diagram through a target extraction network, generating a preselected frame on the target characteristic diagram by an RPN network, generating a labeled characteristic diagram by an ROI Align layer, and finally inputting the labeled characteristic diagram into an FCN network to obtain an insulating film gap detection result, wherein if the insulating film gap in the labeled characteristic diagram is larger than a preset gap error range, the insulating film gap detection result is determined to be a gap width; and if the gap of the insulating film in the labeled characteristic diagram is smaller than a preset gap error range, determining that the gap detection result of the insulating film is narrow.

105. Inputting an insulating film image into a preset insulating film abnormity detection model to carry out insulating film abnormity detection, and obtaining an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulation film breakage, insulation film loss, and insulation film wrinkle;

it should be noted that, when the insulating film is abnormal, usually because the insulating film has an insulating film body defect and thus does not meet the preset condition, in the embodiment of the present invention, the server inputs the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality type analysis, so as to obtain an insulating film abnormality type, and further obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulating film breakage, insulating film loss, and insulating film wrinkling.

106. And extracting the defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating the characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values, and generating the lithium battery defect detection result corresponding to the lithium battery to be detected.

In the embodiment of the invention, the server determines the problems of the positive electrode and the negative electrode according to the positive and negative electrode deviation result, judges whether short circuit is possible according to the insulating film gap detection result and the insulating film abnormality detection result, and further generates a corresponding defect detection result. In addition, the defect regions in the positive and negative electrode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result are extracted, the defect regions are subjected to defect coding calculation to obtain characteristic values, the defect regions are classified and stored according to the characteristic values to generate lithium battery defect detection results corresponding to the lithium battery to be detected, the defect coding calculation is to obtain the defect type and the characteristic value corresponding to each defect region by matching the defect types in the positive and negative electrode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result with a preset defect type table, and the defect types and the characteristic values corresponding to the defect regions are used as lithium battery defect detection results and are output.

In the embodiment of the invention, a first detection image of a lithium battery to be detected is acquired based on a preset image acquisition terminal, and a second detection image of the lithium battery to be detected is acquired, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises an insulating film of the lithium battery to be detected; denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection to obtain an insulating film gap detection result; inputting the insulating film image into a preset insulating film abnormity detection model to carry out insulating film abnormity detection to obtain an insulating film abnormity detection result; and generating a lithium battery defect detection result corresponding to the lithium battery to be detected according to the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result. According to the invention, three deep learning models are constructed in advance, and the defects of the insulation film, the insulation film gap and the positive and negative electrodes of the lithium battery are detected in a targeted manner through the three deep learning models, so that the accuracy of the defect detection of the lithium battery is improved.

Referring to fig. 2, another embodiment of the method for detecting defects of a lithium battery based on an image according to an embodiment of the present invention includes:

201. setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to the preset insulation film point location information, wherein the first detection image comprises an anode and a cathode of the lithium battery to be detected, and the second detection image comprises an insulation film of the lithium battery to be detected;

specifically, in this embodiment, the specific implementation of step 201 is similar to that of step 101, and is not described herein again.

202. Denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image;

specifically, the server inputs the first detection image into a preset EDSR model for noise removal to obtain a first high-definition detection image; cutting the first high-definition detection image according to a preset image size to obtain positive and negative electrode images; inputting the second detection image into an EDSR model for noise removal to obtain a second high-definition detection image; performing area detection on the second high-definition detection image to obtain an insulating film area and an insulating film gap area; and performing region segmentation on the second high-definition detection image according to the insulating film region and the insulating film gap region to obtain an insulating film gap image and an insulating film image.

The method comprises the steps of obtaining the cutting scale of a detection image according to the regional characteristics of the detection image, partitioning the detection image according to the cutting scale of the detection image, and setting the overlapping region among a plurality of partitioned image blocks. The method comprises the steps of performing frequency domain transformation on a plurality of detection image blocks to obtain a plurality of frequency domain image blocks, determining fusion weights of the plurality of detection image frequency domain image blocks during fusion according to inter-block differences, performing frequency domain filtering and inverse frequency domain transformation on the fused images to obtain transformed images, and performing slicing and merging on the transformed images according to the overlapping state of the overlapping regions of the detection images to obtain noise-reduced detection images.

203. Inputting a preset positive and negative electrode deviation detection model with a positive and negative electrode image, wherein the positive and negative electrode deviation detection model comprises: a Darknet-53 network, a Batch nonilization layer, a hidden layer and an output layer;

204. inputting the positive and negative electrode images into a Darknet-53 network for feature extraction to obtain positive and negative electrode feature images;

205. inputting the positive and negative characteristic images into a Batch non-normalization layer for Batch normalization processing to obtain normalized characteristic images;

206. inputting the normalized feature image into a hidden layer to perform feature superposition to obtain a fine-grained feature image;

207. inputting the feature graph with fine granularity into an output layer to predict positive and negative deviation points, and obtaining a predicted image of the positive and negative pixel points;

specifically, the Darknet-53 network inputs the positive and negative electrode images into the network, extracts the corresponding positive and negative electrode characteristic images, and then deepens, refines and corrects the characteristics of the positive and negative electrode characteristic images through each convolution. And the Batch normalization processing is carried out on the positive and negative characteristic diagrams generated from each convolution layer in the Darknet-53 network through the Batch normalization layer by the Batch normalization layer, and the data of the characteristic diagrams are subjected to normalization processing for improving the convergence of the model. And overlapping the corresponding pooled or convolved feature maps, further expanding the resolution of the features, and refining the fine granularity of the features. The feature number of the feature map is increased after each convolution, the channel number is increased after each pooling, and the fine granularity is refined, so that the feature depth of the feature map is increased. After each convolution, calculating the offset of the prior frame by comparing the central coordinates and the width and height sizes of the prior frame and the feature picture frame, correspondingly adjusting the central coordinates and the area sizes of the prior frame, gradually optimizing the target detection accuracy of the feature map, and finally inputting the normalized feature image into a hidden layer by a server for feature superposition to obtain a fine-grained feature map; and inputting the feature map with fine granularity into an output layer to predict positive and negative deviation points, so as to obtain a predicted image of the positive and negative pixel points.

208. Generating positive and negative electrode deviation detection results corresponding to the positive and negative electrode images according to the positive and negative electrode pixel point predicted images, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation;

specifically, the server extracts a plurality of positive electrode pixel points and a plurality of negative electrode pixel points in the positive electrode pixel point and negative electrode pixel point predicted image; and carrying out pixel-level classification comparison on the plurality of positive electrode pixel points and the plurality of negative electrode pixel points to obtain a positive and negative electrode deviation detection result.

After the server extracts a plurality of positive pixel points and a plurality of negative pixel points in the positive and negative pixel point prediction image, the server establishes a plane rectangular coordinate system on the plane where the current picture is located, performs coordinate analysis on the plurality of positive pixel points and the plurality of negative pixel points, determines coordinate values of the pixel points, and performs pixel-level classification comparison on the plurality of positive pixel points and the plurality of negative pixel points through the coordinate values of the pixel points to obtain a positive and negative deviation detection result.

209. Inputting an insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises the following steps: wide and narrow gaps;

specifically, the server inputs an insulating film gap image into a preset insulating film gap detection model, wherein the insulating film gap detection model includes: the method comprises the following steps of (1) extracting a target feature, an RPN, an ROI Align layer and an FCN; inputting the insulating film gap image into a target feature extraction network for feature extraction to obtain an insulating film gap feature map; inputting the insulating film gap characteristic diagram into an RPN network, and generating a preselected frame corresponding to the insulating film gap characteristic diagram according to preset anchor frame information through the RPN network; inputting the preselected frame and the insulating film gap characteristic diagram into an ROI (region of interest) alignment layer, performing characteristic fusion on the preselected frame and the insulating film gap characteristic diagram through the ROI alignment layer, and performing segmentation and end point pooling on the preselected frame to obtain an insulating film gap characteristic labeling diagram; inputting the insulation film gap characteristic labeling diagram into an FCN network, predicting each pixel point of the insulation film gap characteristic labeling diagram through the FCN network, obtaining and outputting a prediction result of the insulation film gap pixel points; and generating an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result.

Note that RPN (region pro-potential Network) and "region pro-potential" are "region selection". In the RPN network, preset anchor frame information is acquired, whether an identification target is included in the anchor frame is judged, if yes, the anchor frame is reserved, and position adjustment is performed on the anchor frame to obtain a preselected frame of the insulating film gap image. And then fusing the target with the feature map and the preselection frame through an ROI (Region of Interest alignment) layer to obtain a labeled feature map. ROI Align is a region-of-interest matching method for solving the problem of region mismatching caused by two quantization in ROI Pooling operation. And then obtaining a preselected frame of the labeled characteristic diagram, a probability value and a mask of each pixel point through an FCN (full convolution network), and outputting the preselected frame, the probability value and the mask as an identification result. The full convolution network replaces the traditional full connection layer, and the final server obtains and outputs the prediction result of the insulation film gap pixel points; and generating an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result.

Specifically, the server measures the gap distance of the prediction result of the insulating film gap pixel point to obtain the target gap width; comparing the target gap width with a preset gap threshold value to obtain a gap comparison result; and generating an insulating film gap detection result according to the gap comparison result.

In the embodiment of the present invention, when the insulating film is abnormal, the server inputs an insulating film image into a preset insulating film abnormality detection model to perform gap distance measurement, obtain a corresponding distance, perform abnormality analysis according to the distance, and further obtain an insulating film abnormality detection result, in the embodiment of the present invention, the server establishes a planar rectangular coordinate system for a plane where insulating film gap pixel points are located, determines coordinate values of each insulating film gap pixel point, further performs gap width calculation according to a two-point distance formula, obtains a target gap width, and compares the target gap width with a preset gap threshold to obtain a gap comparison result; and generating an insulating film gap detection result according to the gap comparison result.

210. Inputting the insulating film image into a preset insulating film abnormity detection model to carry out insulating film abnormity detection, and obtaining an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulating film breakage, insulating film loss, and insulating film wrinkle;

it should be noted that, when the insulating film is abnormal, usually because the insulating film has a defect in the insulating film body and thus the insulating film does not meet the preset condition, in the embodiment of the present invention, the server inputs the image of the insulating film into a preset insulating film abnormality detection model to perform insulating film abnormality type analysis, so as to obtain an insulating film abnormality type and further obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulating film breakage, insulating film loss, and insulating film wrinkles. Specifically, the server inputs the insulating film image into a preset insulating film abnormality detection model, wherein the insulating film abnormality detection model includes: a convolutional network, a plurality of residual error networks, and an output layer; performing convolution operation on the insulating film image through a convolution network to obtain a high-dimensional insulating film image; inputting the high-dimensional insulating film image into a plurality of residual error networks for residual error processing to obtain a characteristic insulating film image; and inputting the characteristic insulating film image into the output layer to perform insulating film pixel point detection, generating and outputting an insulating film abnormity detection result.

After the image of the insulating film image is input, convolution, activation and residual scaling are firstly carried out, the high-frequency characteristics of the image are learned, and then upsampling (corresponding to downsampling pooling, and upsampling is a step of increasing resolution) is carried out to carry out final ultra-definition reconstruction of the image. The convolution network is used for extracting the features of an image, the weight of a convolution kernel can be learned, the convolution operation can break through the limitation of a traditional filter, the desired features are extracted according to an objective function, the convolution operation is carried out on the insulating film image through the convolution network to obtain a high-dimensional insulating film image, the high-dimensional insulating film image is input into a plurality of residual error networks to be subjected to residual error processing to obtain a feature insulating film image, finally, the server carries out pixel point analysis and detection on the feature insulating film image to judge whether an abnormal analysis result exists, and it needs to be noted that the abnormal analysis result generally comprises abnormal results such as unqualified insulating film thickness, overlarge insulating film gap distance and the like.

211. And extracting the defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating the characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values, and generating the lithium battery defect detection result corresponding to the lithium battery to be detected.

Specifically, in this embodiment, the specific implementation of step 211 is similar to that of step 106, and is not described herein again.

In the embodiment of the invention, a first detection image of a lithium battery to be detected is acquired based on a preset image acquisition terminal, and a second detection image of the lithium battery to be detected is acquired, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises an insulating film of the lithium battery to be detected; denoising and cutting the first detection image to obtain positive and negative images, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection to obtain positive and negative electrode deviation detection results; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection to obtain an insulating film gap detection result; inputting the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection to obtain an insulating film abnormality detection result; and generating a lithium battery defect detection result corresponding to the lithium battery to be detected according to the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result. According to the invention, three deep learning models are constructed in advance, and the defects of the insulation film, the insulation film gap and the positive and negative electrodes of the lithium battery are detected in a targeted manner through the three deep learning models, so that the accuracy of the defect detection of the lithium battery is improved.

In the above description of the image-based lithium battery defect detection method in the embodiment of the present invention, an image-based lithium battery defect detection apparatus in the embodiment of the present invention is described below with reference to fig. 3, and an embodiment of the image-based lithium battery defect detection apparatus in the embodiment of the present invention includes:

the acquisition module 301 is configured to set an image acquisition parameter of an image acquisition terminal, acquire a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset positive and negative electrode point location information, and acquire a second detection image of the lithium battery to be detected according to preset insulation film point location information, where the first detection image includes positive and negative electrodes of the lithium battery to be detected, and the second detection image includes an insulation film of the lithium battery to be detected;

a processing module 302, configured to perform noise reduction and clipping on the first detection image to obtain positive and negative images, and perform noise reduction and image segmentation on the second detection image to obtain an insulating film gap image and an insulating film image;

a deviation detection module 303, configured to input the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, so as to obtain positive and negative electrode deviation detection results, where the positive and negative electrode deviation detection results include: large deviation and small deviation;

a gap detection module 304, configured to input the insulating film gap image into a preset insulating film gap detection model for performing insulating film gap detection, so as to obtain an insulating film gap detection result, where the insulating film gap detection result includes: wide and narrow gaps;

an insulating film detection module 305, configured to input the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection, so as to obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulating film breakage, insulating film loss, and insulating film wrinkle;

the generating module 306 is configured to extract a defect region in the positive and negative electrode deviation detection result, the insulation film gap detection result, and the insulation film abnormality detection result, calculate a feature value of the defect region, classify and store the defect region according to the feature value, and generate a lithium battery defect detection result corresponding to the lithium battery to be detected.

In the embodiment of the invention, image acquisition parameters of an image acquisition terminal are set, a first detection image of a lithium battery to be detected is acquired based on the image acquisition terminal and preset anode and cathode point location information, and a second detection image of the lithium battery to be detected is acquired according to the preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected; denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, and obtaining positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: the gap is wide and narrow; inputting the insulating film image into a preset insulating film abnormity detection model for carrying out insulating film abnormity detection to obtain an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulating film breakage, insulating film loss, and insulating film wrinkle; and extracting defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values, and generating a lithium battery defect detection result corresponding to the lithium battery to be detected. According to the invention, three deep learning models are constructed in advance, and the defects of the insulating film, the insulating film gap and the positive and negative electrodes of the lithium battery are detected in a targeted manner through the three deep learning models, so that the accuracy of the defect detection of the lithium battery is improved.

Referring to fig. 4, another embodiment of the image-based lithium battery defect detecting apparatus according to the embodiment of the present invention includes:

a processing module 302, configured to perform noise reduction and cropping on the first detection image to obtain positive and negative images, and perform noise reduction and image segmentation on the second detection image to obtain an insulating film gap image and an insulating film image;

a deviation detecting module 303, configured to input the positive and negative electrode images into a preset positive and negative deviation detecting model for positive and negative deviation detection, so as to obtain a positive and negative deviation detecting result, where the positive and negative deviation detecting result includes: the deviation is large and small;

an insulating film detection module 305, configured to input the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection, so as to obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulation film breakage, insulation film loss, and insulation film wrinkle;

Optionally, the processing module 302 is specifically configured to: inputting the first detection image into a preset EDSR model for noise removal to obtain a first high-definition detection image; cutting the first high-definition detection image according to a preset image size to obtain positive and negative electrode images; inputting the second detection image into the EDSR model for noise removal to obtain a second high-definition detection image; carrying out region detection on the second high-definition detection image to obtain an insulating film region and an insulating film gap region; and carrying out region segmentation on the second high-definition detection image according to the insulating film region and the insulating film gap region to obtain an insulating film gap image and an insulating film image.

Optionally, the deviation detecting module 303 is specifically configured to:

a deviation detecting unit 3031, configured to input the positive and negative electrode images into a preset positive and negative electrode deviation detection model, where the positive and negative electrode deviation detection model includes: a Darknet-53 network, a Batch nonillization layer, a hidden layer, and an output layer; inputting the positive and negative electrode images into the Darknet-53 network for feature extraction to obtain positive and negative electrode feature images; inputting the positive and negative characteristic images into the Batch nonallimation layer for Batch normalization processing to obtain normalized characteristic images; inputting the normalized feature image into the hidden layer for feature superposition to obtain a fine-grained feature image; inputting the fine-grained characteristic diagram into the output layer to predict positive and negative electrode deviation points to obtain a positive and negative electrode pixel point predicted image;

a generating unit 3032, configured to generate a positive-negative deviation detection result corresponding to the positive-negative image according to the positive-negative pixel point prediction image, where the positive-negative deviation detection result includes: the deviation is large and the deviation is small.

Optionally, the generating unit 3032 is specifically configured to: extracting a plurality of positive electrode pixel points and a plurality of negative electrode pixel points in the positive electrode pixel point and negative electrode pixel point predicted image; and carrying out pixel-level classification comparison on the plurality of positive electrode pixel points and the plurality of negative electrode pixel points to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and the deviation is small.

Optionally, the gap detection module 304 is specifically configured to:

a gap detection unit 3041 for inputting the insulating film gap image into a preset insulating film gap detection model, wherein the insulating film gap detection model includes: the method comprises the steps of extracting a target feature, an RPN, an ROI Align layer and an FCN; inputting the insulating film gap image into the target feature extraction network for feature extraction to obtain an insulating film gap feature map; inputting the insulating film gap characteristic diagram into the RPN network, and generating a preselected frame corresponding to the insulating film gap characteristic diagram according to preset anchor frame information through the RPN network; inputting the preselected frame and the insulating film gap characteristic diagram into the ROI Align layer, performing characteristic fusion on the preselected frame and the insulating film gap characteristic diagram through the ROI Align layer, and performing segmentation and endpoint pooling on the preselected frame to obtain an insulating film gap characteristic labeling diagram; inputting the insulation film gap characteristic labeling diagram into the FCN, predicting each pixel point of the insulation film gap characteristic labeling diagram through the FCN, obtaining and outputting an insulation film gap pixel point prediction result;

a result generating unit 3042, configured to generate an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result, where the insulating film gap detection result includes: the gap is wide and the gap is narrow.

Optionally, the result generating unit 3042 is specifically configured to: measuring the gap distance of the prediction result of the insulating film gap pixel point to obtain the target gap width; comparing the target gap width with a preset gap threshold value to obtain a gap comparison result; generating an insulation film gap detection result according to the insulation film gap pixel point prediction result and the gap comparison result, wherein the insulation film gap detection result comprises: the gap is wide and the gap is narrow.

Optionally, the insulating film detection module 305 is specifically configured to: inputting the insulating film image into a preset insulating film abnormality detection model, wherein the insulating film abnormality detection model includes: a convolutional network, a plurality of residual error networks, and an output layer; performing convolution operation on the insulating film image through the convolution network to obtain a high-dimensional insulating film image; inputting the high-dimensional insulating film image into the residual error networks to carry out residual error processing to obtain a characteristic insulating film image; inputting the characteristic insulating film image into the output layer to perform insulating film pixel point detection, generating and outputting an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises: insulating film breakage, insulating film loss, and insulating film wrinkling.

In the embodiment of the invention, image acquisition parameters of an image acquisition terminal are set, a first detection image of a lithium battery to be detected is acquired based on the image acquisition terminal and preset anode and cathode point location information, and a second detection image of the lithium battery to be detected is acquired according to preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises an insulation film of the lithium battery to be detected; denoising and cutting the first detection image to obtain a positive electrode image and a negative electrode image, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image; inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, and obtaining positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation; inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: the gap is wide and narrow; inputting the insulating film image into a preset insulating film abnormity detection model for carrying out insulating film abnormity detection to obtain an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulation film breakage, insulation film loss, and insulation film wrinkle; and extracting defect areas in the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result, calculating characteristic values of the defect areas, classifying and storing the defect areas according to the characteristic values, and generating a lithium battery defect detection result corresponding to the lithium battery to be detected. According to the invention, three deep learning models are constructed in advance, and the defects of the insulation film, the insulation film gap and the positive and negative electrodes of the lithium battery are detected in a targeted manner through the three deep learning models, so that the accuracy of the defect detection of the lithium battery is improved.

Fig. 3 and 4 describe the image-based lithium battery defect detection apparatus in the embodiment of the present invention in detail from the perspective of a modular functional entity, and the image-based lithium battery defect detection apparatus in the embodiment of the present invention is described in detail from the perspective of hardware processing.

Fig. 5 is a schematic structural diagram of an image-based lithium battery defect detecting apparatus 500 according to an embodiment of the present invention, where the image-based lithium battery defect detecting apparatus 500 may generate relatively large differences due to different configurations or performances, and may include one or more processors (CPUs) 510 (e.g., one or more processors) and a memory 520, and one or more storage media 530 (e.g., one or more mass storage devices) storing applications 533 or data 532. Memory 520 and storage media 530 may be, among other things, transient storage or persistent storage. The program stored in the storage medium 530 may include one or more modules (not shown), and each module may include a series of instruction operations for the image-based lithium battery defect detecting apparatus 500. Still further, the processor 510 may be configured to communicate with the storage medium 530, and execute a series of instruction operations in the storage medium 530 on the image-based lithium battery defect detecting apparatus 500.

The image-based lithium battery defect detection apparatus 500 may further include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input-output interfaces 560, and/or one or more operating systems 531, such as Windows server, Mac OS X, Unix, Linux, FreeBSD, and the like. Those skilled in the art will appreciate that the configuration of the image-based lithium battery defect detecting apparatus shown in fig. 5 does not constitute a limitation of the image-based lithium battery defect detecting apparatus, and may include more or less components than those shown, or some components in combination, or a different arrangement of components.

The invention also provides an image-based lithium battery defect detection device, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the image-based lithium battery defect detection method in the above embodiments.

The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium, and which may also be a volatile computer-readable storage medium, having stored therein instructions, which, when executed on a computer, cause the computer to perform the steps of the image-based lithium battery defect detection method.

Further, the computer-readable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created according to the use of the blockchain node, and the like.

The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a portable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, an optical disk, or other various media capable of storing program codes.

The above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. The image-based lithium battery defect detection method is characterized by comprising the following steps of:

setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to the preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected;

denoising and cutting the first detection image to obtain positive and negative images, and denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image;

inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, and obtaining positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation;

inputting the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, and obtaining an insulating film gap detection result, wherein the insulating film gap detection result comprises: the gap is wide and narrow;

inputting the insulating film image into a preset insulating film abnormity detection model for carrying out insulating film abnormity detection to obtain an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises the following steps: insulating film breakage, insulating film loss, and insulating film wrinkle;

and extracting defect areas in the positive and negative electrode deviation detection result, the insulation film gap detection result and the insulation film abnormity detection result, calculating characteristic values of the defect areas, classifying and storing the defect areas according to the characteristic values, and generating a lithium battery defect detection result corresponding to the lithium battery to be detected.

2. The image-based lithium battery defect detection method of claim 1, wherein the denoising and cutting the first detection image to obtain positive and negative electrode images, and the denoising and image segmentation processing the second detection image to obtain an insulating film gap image and an insulating film image comprises:

inputting the first detection image into a preset EDSR model for noise removal to obtain a first high-definition detection image;

cutting the first high-definition detection image according to a preset image size to obtain positive and negative electrode images;

inputting the second detection image into the EDSR model for noise removal to obtain a second high-definition detection image;

carrying out region detection on the second high-definition detection image to obtain an insulating film region and an insulating film gap region;

and carrying out region segmentation on the second high-definition detection image according to the insulating film region and the insulating film gap region to obtain an insulating film gap image and an insulating film image.

3. The image-based lithium battery defect detection method of claim 1, wherein the positive and negative electrode images are input into a preset positive and negative electrode deviation detection model for positive and negative electrode deviation detection, so as to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and small, including:

inputting the positive and negative electrode images into a preset positive and negative electrode deviation detection model, wherein the positive and negative electrode deviation detection model comprises: a Darknet-53 network, a Batch nonillization layer, a hidden layer, and an output layer;

inputting the positive and negative electrode images into the Darknet-53 network for feature extraction to obtain positive and negative electrode feature images;

inputting the positive and negative characteristic images into the Batch nonallimation layer for Batch normalization processing to obtain normalized characteristic images;

inputting the normalized feature image into the hidden layer for feature superposition to obtain a fine-grained feature image;

inputting the feature map with the fine granularity into the output layer to predict positive and negative deviation points, so as to obtain a predicted image of the positive and negative pixel points;

generating positive and negative electrode deviation detection results corresponding to the positive and negative electrode images according to the positive and negative electrode pixel point predicted images, wherein the positive and negative electrode deviation detection results comprise: large deviation and small deviation.

4. The image-based lithium battery defect detection method of claim 3, wherein the positive and negative electrode deviation detection results corresponding to the positive and negative electrode images are generated according to the positive and negative electrode pixel point predicted image, wherein the positive and negative electrode deviation detection results include: the deviation is big and little, including:

extracting a plurality of positive pixel points and a plurality of negative pixel points in the positive and negative pixel point predicted image;

and carrying out pixel-level classification comparison on the plurality of positive electrode pixel points and the plurality of negative electrode pixel points to obtain positive and negative electrode deviation detection results, wherein the positive and negative electrode deviation detection results comprise: the deviation is large and the deviation is small.

5. The image-based lithium battery defect detection method of claim 1, wherein the insulation film gap image is input into a preset insulation film gap detection model for insulation film gap detection, so as to obtain an insulation film gap detection result, wherein the insulation film gap detection result comprises: the gap is wide and narrow, including:

inputting the insulation film gap image into a preset insulation film gap detection model, wherein the insulation film gap detection model comprises: the method comprises the following steps of (1) extracting a target feature, an RPN, an ROI Align layer and an FCN;

inputting the insulating film gap image into the target feature extraction network for feature extraction to obtain an insulating film gap feature map;

inputting the insulating film gap characteristic diagram into the RPN network, and generating a preselected frame corresponding to the insulating film gap characteristic diagram according to preset anchor frame information through the RPN network;

inputting the preselected frame and the insulating film gap characteristic diagram into the ROI Align layer, performing characteristic fusion on the preselected frame and the insulating film gap characteristic diagram through the ROI Align layer, and performing segmentation and endpoint pooling on the preselected frame to obtain an insulating film gap characteristic labeled diagram;

inputting the insulation film gap characteristic labeling diagram into the FCN, predicting each pixel point of the insulation film gap characteristic labeling diagram through the FCN, obtaining an insulation film gap pixel point prediction result and outputting the insulation film gap pixel point prediction result;

generating an insulating film gap detection result corresponding to the insulating film gap image according to the insulating film gap pixel point prediction result, wherein the insulating film gap detection result comprises: the gap is wide and the gap is narrow.

6. The image-based lithium battery defect detection method of claim 5, wherein the insulation film gap detection result corresponding to the insulation film gap image is generated according to the insulation film gap pixel point prediction result, wherein the insulation film gap detection result comprises: the gap width and the gap width include:

measuring the gap distance of the prediction result of the insulating film gap pixel point to obtain the target gap width;

comparing the target gap width with a preset gap threshold value to obtain a gap comparison result;

generating an insulation film gap detection result according to the insulation film gap pixel point prediction result and the gap comparison result, wherein the insulation film gap detection result comprises: the gap is wide and the gap is narrow.

7. The image-based lithium battery defect detection method of any one of claims 1-6, wherein the insulation film image is input into a preset insulation film abnormality detection model for insulation film abnormality detection to obtain an insulation film abnormality detection result, wherein the insulation film abnormality detection result comprises: insulating film breakage, insulating film loss, and insulating film wrinkle, including:

inputting the insulating film image into a preset insulating film abnormality detection model, wherein the insulating film abnormality detection model comprises: a convolutional network, a plurality of residual error networks, and an output layer;

performing convolution operation on the insulating film image through the convolution network to obtain a high-dimensional insulating film image;

inputting the high-dimensional insulating film image into the residual error networks to carry out residual error processing to obtain a characteristic insulating film image;

inputting the characteristic insulating film image into the output layer to perform insulating film pixel point detection, generating and outputting an insulating film abnormity detection result, wherein the insulating film abnormity detection result comprises: insulating film breakage, insulating film loss, and insulating film wrinkles.

8. An image-based lithium battery defect detection device, characterized in that the image-based lithium battery defect detection device comprises:

the system comprises an acquisition module, a detection module and a processing module, wherein the acquisition module is used for setting image acquisition parameters of an image acquisition terminal, acquiring a first detection image of a lithium battery to be detected based on the image acquisition terminal and preset anode and cathode point location information, and acquiring a second detection image of the lithium battery to be detected according to preset insulation film point location information, wherein the first detection image comprises the anode and the cathode of the lithium battery to be detected, and the second detection image comprises the insulation film of the lithium battery to be detected;

the processing module is used for carrying out noise reduction and cutting on the first detection image to obtain positive and negative electrode images, and carrying out noise reduction and image segmentation processing on the second detection image to obtain an insulating film gap image and an insulating film image;

a deviation detection module, configured to input the positive and negative electrode images into a preset positive and negative electrode deviation detection model to perform positive and negative electrode deviation detection, so as to obtain positive and negative electrode deviation detection results, where the positive and negative electrode deviation detection results include: large deviation and small deviation;

a gap detection module, configured to input the insulating film gap image into a preset insulating film gap detection model to perform insulating film gap detection, so as to obtain an insulating film gap detection result, where the insulating film gap detection result includes: the gap is wide and narrow;

an insulating film detection module, configured to input the insulating film image into a preset insulating film abnormality detection model to perform insulating film abnormality detection, so as to obtain an insulating film abnormality detection result, where the insulating film abnormality detection result includes: insulation film breakage, insulation film loss, and insulation film wrinkle;

and the generating module is used for extracting the defect regions in the anode and cathode deviation detection result, the insulating film gap detection result and the insulating film abnormity detection result, calculating the characteristic values of the defect regions, classifying and storing the defect regions according to the characteristic values and generating the lithium battery defect detection result corresponding to the lithium battery to be detected.

9. An image-based lithium battery defect detecting apparatus, characterized in that the image-based lithium battery defect detecting apparatus comprises: a memory and at least one processor, the memory having instructions stored therein;

the at least one processor invokes the instructions in the memory to cause the image-based lithium battery defect detection device to perform the image-based lithium battery defect detection method of any of claims 1-7.

10. A computer readable storage medium having instructions stored thereon, wherein the instructions, when executed by a processor, implement the image-based lithium battery defect detection method according to any one of claims 1 to 7.