A battery top cover defect detection system and method based on image analysis

Abstract

The present invention discloses a battery top cover defect detection system and method based on image analysis, which relate to the field of battery top cover defect detection, comprising: a robotic arm module, used for continuously grabbing the battery top cover to be detected; a rotating module, used for carrying the robotic arm module to rotate at a predetermined speed; an acquisition module, used for acquiring image data of the battery top cover in a rotating state and storing the battery top cover image data; a construction module, used for uploading standard structural parameters of the battery top cover and constructing a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover; the present invention abandons the traditional single detection mode and adopts an innovative method of multiple grabbing combined with rotation acquisition to acquire battery top cover images from different positions and angles, obtain appearance information in all directions, and greatly improve the detection accuracy and comprehensiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of battery top cover defect detection, and in particular to a battery top cover defect detection system and method based on image analysis. Background Art

[0002] The battery cover is a critical component of the battery, located at the top of the cell, providing sealing, protection, and connectivity. Made of metal or composite materials, it offers insulation and corrosion resistance, shielding it from external impurities such as moisture and dust, and preventing internal short circuits. The battery cover also integrates structures such as terminals and a safety valve to facilitate power transmission and internal pressure relief, playing a crucial role in ensuring safe and stable battery operation.

[0003] The invention patent application with publication number 202210778944.6 discloses a battery top cover defect detection method, including: obtaining the image of the battery top cover to be tested by partitioning to obtain images corresponding to different areas of the battery top cover to be tested; inputting the images corresponding to different areas into the corresponding defect detection composite model at the same time to perform parallel detection on the images corresponding to different areas to obtain defect detection sub-results corresponding to each of the areas; determining the defect detection results of the battery top cover to be tested based on the defect detection sub-results corresponding to each of the areas. This application aims to solve the problem that "however, there are many types of power batteries with different sizes and they are updated quickly. In the process of top cover welding defect detection, due to the large differences in the lengths of the battery top covers, complete and clear image data cannot be collected, resulting in the existing technology being incompatible with the top cover detection of different power batteries."

[0004] However, in the application process of existing technologies for visually detecting defects in battery top covers, the collected battery top cover images are often unable to escape the nature of the static body of the battery top cover, which hinders the optimization of the detection accuracy of this detection technology.

[0005] Therefore, a battery top cover defect detection system and method based on image analysis are proposed. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides a battery top cover defect detection system and method based on image analysis, which can effectively solve the problems of the prior art.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] The present invention discloses a battery top cover defect detection system based on image analysis, comprising: a robotic arm module for continuously grabbing a battery top cover to be detected; a rotating module for carrying the robotic arm module to rotate at a predetermined speed; an acquisition module for acquiring image data of the battery top cover in a rotating state and storing the battery top cover image data; a construction module for uploading standard structural parameters of the battery top cover and constructing a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover; a picking module for receiving the standard three-dimensional model of the battery top cover constructed in the construction module and picking up a reference image in the standard three-dimensional model of the battery top cover; a comparison module for comparing the battery top cover image data stored in the acquisition module with the reference image picked up by the picking module and identifying the difference between the two groups of images; a determination module for setting a defect determination threshold, synchronously receiving a difference recognition result in the comparison module, and determining whether the battery top cover from which the difference recognition result is derived is defective based on the comparison of the difference recognition result with the defect determination threshold.

[0009] Furthermore, the gripping end of the robotic arm module is provided with an electric suction cup, and the robotic arm module continuously grips the battery top cover at different positions on the surface of the battery top cover through the electric suction cup. After each gripping of the battery top cover, the rotating module carries the robotic arm module to rotate once at a predetermined speed, and the duration of a single rotation is greater than the time required for the acquisition module to operate once;

[0010] Among them, the robotic arm module grabs the same battery top cover to be tested no less than three times, that is, the rotation module and the acquisition module synchronize the robotic arm module to run the same number of times, the acquisition module is integrated by the camera device, and the shutter speed of the camera device is lower than the speed at which the rotation module carries the robotic arm module, that is, the time required for the camera device shutter to open once is greater than the time it takes for the rotation module to carry the robotic arm module to rotate one circle.

[0011] Furthermore, the robotic arm module and the rotary module are painted with a designated color that is different from the color of the battery top cover before being put into use;

[0012] The acquisition module is provided with a segmentation unit and a marking unit at the lower level. The segmentation unit is used to receive the battery top cover image data in the rotation state collected by the acquisition module, and segment the battery top cover area image from the battery top cover image data. The marking unit is used to obtain the source posture information of the battery top cover image data in each rotation state, and the posture information when the acquisition module collects the image data, and mark the two sets of posture information on the corresponding battery top cover area image obtained by the segmentation operation of the segmentation unit, and store the marked battery top cover area image as the storage target of the acquisition module;

[0013] Among them, the battery top cover image data in the rotating state is composed of the battery top cover area image and the background image. The background image comes from the robotic arm module and the rotating module. When the segmentation unit segments the battery top cover image data in the rotating state, the battery top cover area image is distinguished in the image data based on the paint colors of the robotic arm module and the rotating module.

[0014] Furthermore, the image data source posture information of the battery top cover in the rotation state is:

[0015] Take any three or more corner points on the battery cover surface as reference points, and use the deployed ranging device to measure the distance from itself to each corner point and its own position information to obtain the position information of each reference point, which is recorded as posture information;

[0016] The posture information when the acquisition module acquires image data is:

[0017] The camera device's camera end position information and angle;

[0018] Among them, the camera end position information and angle of the camera device are all based on the reference point position information of the battery top cover.

[0019] Furthermore, when the picking module picks up the reference image in the standard three-dimensional model of the battery top cover, it synchronizes with the posture information of the image mark of the battery top cover area in the acquisition module, and based on the traversed posture information, intercepts the image of the standard three-dimensional model of the battery top cover in the three-dimensional space where the standard three-dimensional model of the battery top cover constructed by the construction module is located in the same posture and in a rotated state, that is, the reference image;

[0020] After the picking module picks up the reference image, it jumps to the acquisition module to run again, and the acquisition module runs again to collect a static image of the battery top cover. The picking module further picks up a static reference image of the same perspective on the standard three-dimensional model of the battery top cover;

[0021] Among them, the number of acquisition and picking of the static image of the battery top cover and the static reference image is not less than 1, and is customized by the system end user, and is subject to the higher the accuracy requirement of the battery top cover defect detection, the more acquisitions are made, and vice versa, the fewer acquisitions are made.

[0022] Furthermore, when performing the comparison operation, the comparison module uses the battery top cover image data stored in the acquisition module, that is, the battery top cover area image. The image difference recognition logic in the comparison module is expressed as follows:

[0023] ;

[0024] Where: is the image difference; is the total amount of combinations of battery top cover area images and corresponding reference images; is the total number of combinations of the static image of the battery top cover and the corresponding static reference image; is the similarity between the top cover area image of the i-th group of batteries and the corresponding reference image; is the similarity between the static image of the jth group of battery top covers and the corresponding static reference image; 、 is the weight;

[0025] in, Express To find the average, Similarly, The larger the value, the greater the image difference, that is, the greater the difference between the actual structural parameters of the battery top cover and the standard structural parameters of the battery top cover. and Are greater than 0, weight The value follows that the more corner points on the battery cover surface, the higher the weight. The smaller the value, the greater the weight. The larger the value.

[0026] Furthermore, the similarity calculation logic of two images is expressed as:

[0027] China-Israel For example:

[0028] ;

[0029] Where: is the diameter of the top cover area image of the i-th group of batteries and the corresponding reference image; To adjust the index, hour, =-1, otherwise, =1;

[0030] China-Israel For example:

[0031] Perform feature extraction on the static image of the jth group of battery top cover and the corresponding static reference image, and convert them into feature vectors containing geometric features, including: side length, angle, area, and topological features, including: node connection relationship and number of rings and , the vector dimension is x; define the weight vector , used to indicate the importance of each feature dimension, and obeys ;

[0032] but ;

[0033] Where: The eigenvectors are and The cth component of represents the Euclidean distance between two components, is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the sensitivity of the similarity calculation.

[0034] Furthermore, when the determination module determines that a battery top cover has defects, the determination module controls the robot arm module to sort the battery top covers, so that the battery top covers determined to have defects based on system detection are separated and collected from the battery top covers that do not have defects.

[0035] Furthermore, the robotic arm module and the rotating module are electrically interconnected through a medium, the robotic arm module and the rotating module are interactively connected to an acquisition module through a wireless network, the acquisition module is interactively connected to a segmentation unit and a marking unit through a wireless network, the acquisition module is interactively connected to a construction module through a wireless network, the construction module is interactively connected to a picking module and a comparison module through a wireless network, the comparison module is interactively connected to a determination module through a wireless network, and the comparison module is interactively connected to the acquisition module through a wireless network.

[0036] On the other hand, a battery top cover defect detection method based on image analysis includes the following steps:

[0037] Deploy a robotic arm module and a rotation module to grab and rotate the battery top cover to be inspected to collect image data of the battery top cover in a rotating state; segment and obtain a battery top cover area image from the battery top cover image data; upload standard structural parameters of the battery top cover, construct a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover, and intercept an image of the standard three-dimensional model of the battery top cover in the same posture using the posture information of the battery top cover image data at the time of collection on the standard three-dimensional model of the battery top cover, and record it as a reference image; collect static images from the same perspective on the battery top cover and the standard three-dimensional model of the battery top cover again, and record them as a battery top cover static image and a static reference image, respectively; combine the difference comparison between the battery top cover area image and the reference image and the difference comparison between the battery top cover static image and the static reference image to determine whether the battery top cover has defects; and sort the battery top covers according to the determination result of whether the battery top cover has defects.

[0038] Compared with the prior art, the technical solution provided by the present invention has the following beneficial effects:

[0039] The present invention provides a battery top cover defect detection system and method based on image analysis. During execution, the system and method abandon the traditional single detection mode and adopt an innovative method of multiple captures combined with rotational acquisition to collect battery top cover images from different positions and angles, obtain appearance information in all directions, and significantly improve detection accuracy and comprehensiveness.

[0040] At the same time, a 3D model is constructed based on standard structural parameters and reference images are picked up for comparison, providing an accurate reference for detection. The corresponding posture image can be intercepted in 3D space to quickly and accurately identify structural defects in complex-shaped battery top covers.

[0041] In addition, a weighting mechanism is introduced in image difference calculation and defect judgment, which flexibly adjusts the weight according to the number of corner points on the battery top cover surface, significantly enhancing the intelligence and adaptability of detection, effectively reducing the misjudgment rate, and meeting the detection requirements of battery top covers of different models. It builds a solid technical defense line for battery production quality control and improves detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0043] Figure 1 The figure is a schematic diagram of the structure of a battery top cover defect detection system based on image analysis;

[0044] Figure 2 The figure is a flowchart of a battery top cover defect detection method based on image analysis. DETAILED DESCRIPTION

[0045] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0046] The present invention will be further described below with reference to the embodiments.

[0047] Example 1:

[0048] A battery top cover defect detection system based on image analysis in this embodiment, such as Figure 1 As shown, including:

[0049] A robotic arm module, used to continuously grab the top cover of the battery to be tested;

[0050] The grabbing end of the robotic arm module is provided with an electric suction cup. The robotic arm module continuously grabs the battery top cover at different positions on the surface of the battery top cover through the electric suction cup. After each grabbing of the battery top cover, the rotating module carries the robotic arm module to rotate once at a predetermined speed. The duration of a single rotation is greater than the time required for the acquisition module to run once.

[0051] The robotic arm module grasps the same battery top cover to be inspected at least three times, that is, the rotation module and the acquisition module synchronize the robotic arm module to operate the same number of times. The acquisition module is integrated with a camera device, and the shutter speed of the camera device is lower than the speed at which the robotic arm module is carried by the rotation module. That is, the time required for the camera device shutter to open once is greater than the time it takes for the robotic arm module to rotate one circle with the rotation module.

[0052] The rotating module is used to carry the robotic arm module to rotate at a predetermined speed;

[0053] An acquisition module is used to acquire image data of the battery top cover in a rotating state and store the image data of the battery top cover;

[0054] Before being put into use, the robotic arm module and the rotary module are painted with a designated color that is different from the color of the battery cover;

[0055] The acquisition module is provided with a segmentation unit and a marking unit at the lower level. The segmentation unit is used to receive the battery top cover image data in the rotation state collected by the acquisition module, and segment the battery top cover area image from the battery top cover image data. The marking unit is used to obtain the posture information of the battery top cover image data source in each rotation state, and the posture information when the acquisition module collects the image data, and mark the two sets of posture information on the corresponding battery top cover area image obtained by the segmentation operation of the segmentation unit, and store the marked battery top cover area image as the storage target of the acquisition module;

[0056] The image data of the battery top cover in the rotating state is composed of an image of the battery top cover area and a background image. The background image comes from the robotic arm module and the rotating module. When the segmentation unit performs segmentation processing on the image data of the battery top cover in the rotating state, the image of the battery top cover area is distinguished from the image data based on the paint colors of the robotic arm module and the rotating module.

[0057] The image data source posture information of the battery top cover in the rotated state is:

[0058] Take any three or more corner points on the battery cover surface as reference points, and use the deployed ranging device to measure the distance from itself to each corner point and its own position information to obtain the position information of each reference point, which is recorded as posture information;

[0059] The posture information when the acquisition module collects image data is:

[0060] The camera device's camera end position information and angle;

[0061] The camera device's camera end position information and angle are referenced to the reference point position information of the battery cover;

[0062] A construction module is used to upload standard structural parameters of the battery top cover and build a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover;

[0063] A picking module is used to receive the standard three-dimensional model of the battery top cover constructed in the construction module and pick up a reference image in the standard three-dimensional model of the battery top cover;

[0064] When the picking module picks up the reference image in the standard three-dimensional model of the battery top cover, it synchronously traverses the posture information of the image mark of the battery top cover area in the acquisition module, and based on the traversed posture information, intercepts the image of the standard three-dimensional model of the battery top cover in the three-dimensional space where the standard three-dimensional model of the battery top cover constructed by the construction module is located in the same posture and in a rotated state, which is also the reference image;

[0065] After the picking module picks up the reference image, it jumps to the acquisition module and runs again. The acquisition module runs again to collect a static image of the battery top cover. The picking module further picks up a static reference image of the same perspective on the standard three-dimensional model of the battery top cover;

[0066] The number of acquisition and picking of the static image of the battery top cover and the static reference image is no less than 1 and is customized by the system end user. The higher the requirement for the battery top cover defect detection accuracy, the more acquisitions are required, and vice versa.

[0067] A comparison module is used to compare the battery top cover image data stored in the acquisition module with the reference image picked up by the pickup module to identify the differences between the two sets of images;

[0068] When the comparison module performs the comparison operation, the battery top cover image data stored in the acquisition module is used, that is, the battery top cover area image. The image difference recognition logic in the comparison module is expressed as follows:

[0069] ;

[0070] Where: is the image difference; is the total amount of combinations of battery top cover area images and corresponding reference images; is the total number of combinations of the static image of the battery top cover and the corresponding static reference image; is the similarity between the top cover area image of the i-th group of batteries and the corresponding reference image; is the similarity between the static image of the jth group of battery top covers and the corresponding static reference image; 、 is the weight;

[0071] in, Express To find the average, Similarly, The larger the value, the greater the image difference, that is, the greater the difference between the actual structural parameters of the battery top cover and the standard structural parameters of the battery top cover. and Are greater than 0, weight The value follows that the more corner points on the battery cover surface, the higher the weight. The smaller the value, the greater the weight. The larger the value;

[0072] The similarity calculation logic of two images is expressed as:

[0073] China-Israel For example:

[0074] ;

[0075] Where: is the diameter of the top cover area image of the i-th group of batteries and the corresponding reference image; To adjust the index, hour, =-1, otherwise, =1;

[0076] China-Israel For example:

[0077] Perform feature extraction on the static image of the jth group of battery top cover and the corresponding static reference image, and convert them into feature vectors containing geometric features, including: side length, angle, area, and topological features, including: node connection relationship and number of rings and , the vector dimension is x; define the weight vector , used to indicate the importance of each feature dimension, and obeys ;

[0078] but ;

[0079] Where: The eigenvectors are and The cth component of represents the Euclidean distance between two components, is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the sensitivity of the similarity calculation;

[0080] By calculating through the above logic formula, the calculation logic based on image difference is defined, providing necessary operation data support for the determination module of the system in this embodiment;

[0081] When the determination module determines that a battery top cover has defects, the robot arm module controls the battery top cover to sort the battery top covers, so that the battery top covers determined to have defects based on system detection and the battery top covers without defects are separated and collected;

[0082] A determination module is used to set a defect determination threshold, synchronously receive the difference recognition result from the comparison module, and determine whether the battery top cover from which the difference recognition result is derived is defective based on the comparison between the difference recognition result and the defect determination threshold;

[0083] The robotic arm module and the rotating module are electrically connected through a medium, the robotic arm module and the rotating module are interactively connected to the acquisition module through a wireless network, the acquisition module is interactively connected to the segmentation unit and the marking unit through a wireless network, the acquisition module is interactively connected to the construction module through a wireless network, the construction module is interactively connected to the picking module and the comparison module through a wireless network, the comparison module is interactively connected to the judgment module through a wireless network, and the comparison module is interactively connected to the acquisition module through a wireless network.

[0084] In this embodiment, the robot arm module operates to continuously grab the battery top cover to be inspected, the rotating module synchronously carries the robot arm module to rotate at a predetermined speed, the acquisition module operates to collect the battery top cover image data in the rotating state, stores the battery top cover image data, the segmentation unit synchronously receives the battery top cover image data in the rotating state collected by the acquisition module, segments the battery top cover area image from the battery top cover image data, the marking unit obtains the posture information of the battery top cover image data source in each rotation state and the posture information when the acquisition module collects the image data in real time, and marks the two sets of posture information on the corresponding battery top cover area image obtained by the segmentation operation of the segmentation unit, with the marked battery top cover area image. The domain image is stored as a storage target of the acquisition module, and then the construction module uploads the standard structural parameters of the battery top cover, and constructs a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover. The picking module further receives the standard three-dimensional model of the battery top cover constructed in the construction module, picks up a reference image in the standard three-dimensional model of the battery top cover, and compares the battery top cover image data stored in the acquisition module with the reference image picked up by the picking module through the comparison module to identify the difference between the two groups of images. Finally, the judgment module sets the defect judgment threshold, and synchronously receives the difference recognition result in the comparison module. Based on the comparison of the difference recognition result and the defect judgment threshold, it is determined whether the battery top cover from which the difference recognition result comes is defective.

[0085] Through the operation of the system in the above embodiment, the existing technology of static image collection and detection of battery top covers is eliminated to a certain extent, and the accuracy of battery top cover defect detection based on visual inspection technology is further improved.

[0086] Example 2:

[0087] In terms of specific implementation, based on Example 1, this example refers to Figure 2 The battery top cover defect detection system based on image analysis in Example 1 is further described in detail:

[0088] A battery top cover defect detection method based on image analysis includes the following steps:

[0089] Step 1: Deploy the robotic arm module and the rotation module to grab and rotate the battery top cover to be inspected to collect image data of the battery top cover in a rotating state;

[0090] Step 2: Segment the battery top cover image data to obtain the battery top cover area image;

[0091] Step 3: Upload the standard structural parameters of the battery top cover, build a standard 3D model of the battery top cover based on the standard structural parameters of the battery top cover, and use the posture information of the battery top cover image data at the time of acquisition to intercept the standard 3D model image of the battery top cover in the same posture as the standard 3D model of the battery top cover, and record it as the reference image;

[0092] Step 4: Collect static images from the same viewing angle on the battery top cover and the standard 3D model of the battery top cover again, and record them as the battery top cover static image and static reference image respectively;

[0093] Step 5: Determine whether the battery top cover has defects by comparing the difference between the battery top cover area image and the reference image and the difference between the static image of the battery top cover and the static reference image.

[0094] Step 6: Sort the battery top covers according to the determination result of whether the battery top covers are defective.

[0095] In summary, during the execution of the systems and methods in the above embodiments, the traditional single detection mode is abandoned, and an innovative method of multiple captures combined with rotational acquisition is adopted to collect images of the battery top cover from different positions and angles, obtain appearance information in all directions, and greatly improve the accuracy and comprehensiveness of detection. At the same time, a three-dimensional model is constructed based on standard structural parameters and reference images are picked up for comparison to provide an accurate reference for detection. The corresponding posture image can be intercepted in three-dimensional space, and the structural defects of battery top covers with complex shapes can be quickly and accurately identified. In addition, a weight mechanism is introduced in the image difference calculation and defect judgment, and the weight is flexibly adjusted according to the number of corner points on the battery top cover surface, which significantly enhances the intelligence and adaptability of the detection, effectively reduces the misjudgment rate, meets the detection requirements of battery top covers of different models, builds a solid technical defense line for battery production quality control, and improves detection accuracy.

[0096] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A battery top cover defect detection system based on image analysis, characterized in that: include: A robotic arm module, used to continuously grab the top cover of the battery to be tested; The rotating module is used to carry the robotic arm module to rotate at a predetermined speed; An acquisition module is used to acquire image data of the battery top cover in a rotating state and store the image data of the battery top cover; A construction module is used to upload standard structural parameters of the battery top cover and build a standard three-dimensional model of the battery top cover based on the standard structural parameters of the battery top cover; A picking module is used to receive the standard three-dimensional model of the battery top cover constructed in the construction module and pick up a reference image in the standard three-dimensional model of the battery top cover; A comparison module is used to compare the battery top cover image data stored in the acquisition module with the reference image picked up by the pickup module to identify the differences between the two sets of images; When the comparison module performs the comparison operation, the battery top cover image data stored in the acquisition module is used, that is, the battery top cover area image. The image difference recognition logic in the comparison module is expressed as follows: Where: DIFF is the image difference; n is the total number of combinations of the battery top cover area image and the corresponding reference image; m is the total number of combinations of the battery top cover static image and the corresponding static reference image; SIMM (P i ,Q i ) is the similarity between the image of the top cover area of the i-th group of batteries and the corresponding reference image; SIMM(P j ,Q j ) is the similarity between the static image of the jth battery top cover and the corresponding static reference image; ω, (1-ω) are weights; in, Express To find the average, Similarly, the larger the DIFF is, the greater the image difference is, that is, the greater the difference between the actual structural parameters of the battery top cover and the standard structural parameters of the battery top cover is, ω and (1-ω) are both greater than 0, and the weight ω value follows the following rules: the more corner points on the battery top cover surface, the smaller the weight ω value is, and vice versa, the larger the weight ω value is; The judgment module is used to set a defect judgment threshold, synchronously receive the difference recognition result from the comparison module, and judge whether the battery top cover from which the difference recognition result comes is defective based on the comparison between the difference recognition result and the defect judgment threshold.

2. The battery top cover defect detection system based on image analysis according to claim 1, characterized in that: The grabbing end of the robotic arm module is provided with an electric suction cup, and the robotic arm module continuously grabs the battery top cover at different positions on the surface of the battery top cover through the electric suction cup. After each grabbing of the battery top cover, the rotating module carries the robotic arm module to rotate once at a predetermined speed, and the duration of a single rotation is greater than the time required for the acquisition module to operate once; Among them, the robotic arm module grabs the same battery top cover to be tested no less than three times, that is, the rotation module and the acquisition module synchronize the robotic arm module to run the same number of times, the acquisition module is integrated by the camera device, and the shutter speed of the camera device is lower than the speed at which the rotation module carries the robotic arm module, that is, the time required for the camera device shutter to open once is greater than the time it takes for the rotation module to carry the robotic arm module to rotate one circle.

3. The battery top cover defect detection system based on image analysis according to claim 1, characterized in that: Before being put into use, the robotic arm module and the rotary module are painted with a designated color that is different from the color of the battery top cover; The acquisition module is provided with a segmentation unit and a marking unit at the lower level. The segmentation unit is used to receive the battery top cover image data in the rotation state collected by the acquisition module, and segment the battery top cover area image from the battery top cover image data. The marking unit is used to obtain the source posture information of the battery top cover image data in each rotation state, and the posture information when the acquisition module collects the image data, and mark the two sets of posture information on the corresponding battery top cover area image obtained by the segmentation operation of the segmentation unit, and store the marked battery top cover area image as the storage target of the acquisition module; Among them, the battery top cover image data in the rotating state is composed of the battery top cover area image and the background image. The background image comes from the robotic arm module and the rotating module. When the segmentation unit segments the battery top cover image data in the rotating state, the battery top cover area image is distinguished in the image data based on the paint colors of the robotic arm module and the rotating module.

4. The battery top cover defect detection system based on image analysis according to claim 3, characterized in that: The image data source posture information of the battery top cover in the rotation state is: Take any three or more corner points on the battery cover surface as reference points, and use the deployed ranging device to measure the distance from itself to each corner point and its own position information to obtain the position information of each reference point, which is recorded as posture information; The posture information when the acquisition module acquires image data is: The camera device's camera end position information and angle; Among them, the camera end position information and angle of the camera device are all based on the reference point position information of the battery top cover.

5. The battery top cover defect detection system based on image analysis according to claim 1, characterized in that: When the picking module picks up the reference image in the standard three-dimensional model of the battery top cover, it synchronizes with the posture information of the image mark of the battery top cover area in the acquisition module, and based on the posture information traversed, intercepts the image of the standard three-dimensional model of the battery top cover in the three-dimensional space where the standard three-dimensional model of the battery top cover constructed by the construction module is located in the same posture and in a rotated state, that is, the reference image; After the picking module picks up the reference image, it jumps to the acquisition module to run again, and the acquisition module runs again to collect a static image of the battery top cover. The picking module further picks up a static reference image of the same perspective on the standard three-dimensional model of the battery top cover; Among them, the number of acquisition and picking of the static image of the battery top cover and the static reference image is not less than 1, and is customized by the system end user, and is subject to the higher the accuracy requirement of the battery top cover defect detection, the more acquisitions are made, and vice versa, the fewer acquisitions are made.

6. The battery top cover defect detection system based on image analysis according to claim 1, characterized in that: The similarity calculation logic of two images is expressed as: SIMM(P i ,Q i ) as an example: Where: is the diameter of the top cover area image of the i-th group of batteries and the corresponding reference image; γ is the adjustment index, When , γ=-1, otherwise, γ=1; SIMM(P j ,Q j ) as an example: The feature extraction is performed on the static image of the top cover of the jth group of batteries and the corresponding static reference image, and it is converted into a feature vector v containing geometric features, including: side length, angle, area, and topological features, including: node connection relationship and ring number P and v Q , the vector dimension is x; define the weight vector Used to indicate the importance of each feature dimension and obey but Where: are the eigenvectors v P and v Q The cth component of represents the Euclidean distance between two components, and σ is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the sensitivity of the similarity calculation.

7. The battery top cover defect detection system based on image analysis according to claim 6, characterized in that: When the determination module determines that a battery top cover has defects, the determination module controls the robot arm module to sort the battery top covers, so that the battery top covers determined to have defects based on system detection are separated and collected from the battery top covers that do not have defects.

8. The battery top cover defect detection system based on image analysis according to claim 1, characterized in that: The robotic arm module and the rotating module are electrically interconnected through a medium, the robotic arm module and the rotating module are interconnected with a collection module through a wireless network, the collection module is interactively connected to a segmentation unit and a marking unit through a wireless network, the collection module is interactively connected to a construction module through a wireless network, the construction module is interactively connected to a picking module and a comparison module through a wireless network, the comparison module is interactively connected to a determination module through a wireless network, and the comparison module is interactively connected to the collection module through a wireless network.

9. A battery top cover defect detection method based on image analysis, the method being an implementation of the battery top cover defect detection system based on image analysis according to any one of claims 1 to 8, characterized in that: The following steps are involved: Step 1: Deploy the robotic arm module and the rotation module to grab and rotate the battery top cover to be inspected to collect image data of the battery top cover in a rotating state; Step 2: Segment the battery top cover image data to obtain the battery top cover area image; Step 3: Upload the standard structural parameters of the battery top cover, build a standard 3D model of the battery top cover based on the standard structural parameters of the battery top cover, and use the posture information of the battery top cover image data at the time of acquisition to intercept the standard 3D model image of the battery top cover in the same posture as the standard 3D model of the battery top cover, and record it as the reference image; Step 4: Collect static images from the same viewing angle on the battery top cover and the standard 3D model of the battery top cover again, and record them as the battery top cover static image and static reference image respectively; Step 5: Determine whether the battery top cover has defects by comparing the difference between the battery top cover area image and the reference image and the difference between the static image of the battery top cover and the static reference image. Step 6: Sort the battery top covers according to the determination result of whether the battery top covers are defective.

Images