Defect detection method for battery cell explosion-proof valve, device, medium and system
By acquiring and fusing single-spectral images and three-dimensional point cloud data of different wavelengths, the problem of low detection efficiency of traditional battery cell explosion-proof valves has been solved, achieving efficient and accurate defect identification and type differentiation.
Patent Information
- Application Number
- PCT/CN2023/139616
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-23
- Filing Date
- 2023-12-18
- Publication Date
- 2024-11-28
AI Technical Summary
Traditional methods for detecting defects in battery cell explosion-proof valves are inefficient and cannot accurately distinguish defect types, making it difficult to achieve efficient and accurate detection.
By acquiring multiple single-spectral images of different wavelengths of the target battery cell, fusing them into a multispectral fusion image, and combining it with three-dimensional point cloud data, the defect detection results of the explosion-proof valve, including planar and morphological defects, are determined using computing equipment.
It enables accurate and efficient detection of defects in battery cell explosion-proof valves, and can identify multiple defect types, thus improving detection efficiency and accuracy.
Smart Images

Figure CN2023139616_28112024_PF_FP_ABST
Abstract
Description
Defect detection method, equipment, medium and system for battery cell explosion-proof valve
[0001] This application claims priority to the Chinese patent application filed with the China Patent Office on May 23, 2023, with application number 202310587876.X and invention name “Defect Detection Method, Equipment, Medium and System for Battery Cell Explosion-proof Valve”, the entire contents of which are incorporated by reference into this application. Technical Field
[0002] The present application relates to the field of battery cell detection, and more specifically to a method, device, medium and system for defect detection of a battery cell explosion-proof valve. Background Art
[0003] Battery cell explosion-proof valves usually have multiple functions such as explosion-proof, breathable, waterproof, dust-proof, and oil-proof. Defect detection of battery cell explosion-proof valves is an important part of battery cell safety testing.
[0004] Traditional methods for inspecting battery cell explosion-proof valves include manual inspection or using camera modules to capture images of the battery cell surface and use image recognition to detect defects in the valves. However, manual inspection is inefficient, subjective, and unstable. Camera module inspection typically requires capturing multiple sets of images, which is complex and computationally intensive. Furthermore, existing camera module inspection methods cannot distinguish between the types of defects present in battery cell explosion-proof valves.
[0005] In summary, the disadvantage of the traditional method for detecting defects in battery cell explosion-proof valves is that it is difficult to detect defects in battery cell explosion-proof valves accurately and efficiently. Technical issues
[0006] One of the purposes of the embodiments of the present application is to provide a defect detection method, computing device, storage medium, program product and system for battery cell explosion-proof valves, which can effectively improve the efficiency of battery cell explosion-proof valve defect detection and provide more accurate detection results. Technical Solutions
[0007] To solve the above technical problems, the technical solutions adopted in the embodiments of the present application are:
[0008] According to a first aspect of the present disclosure, a method for defect detection of an explosion-proof valve of a battery cell is provided, comprising: acquiring multiple single-spectrum images of different wavelengths regarding a target battery cell, each single-spectrum image indicating at least an explosion-proof valve region of the target battery cell; fusing the multiple single-spectrum images of different wavelengths to generate a multi-spectral fusion image regarding the target battery cell; and determining a defect detection result regarding the explosion-proof valve of the target battery cell based on the generated multi-spectral fusion image.
[0009] In some embodiments, obtaining single-spectrum images of multiple different wavelengths about the target battery cell includes: controlling the light source of the first detection station to switch to multiple different wavelengths of light in a single exposure time, so that the first image acquisition device can acquire single-spectrum images of multiple different wavelengths about the target battery cell, wherein the station parameters of the first detection station are related to the parameters of the first image acquisition device, the light source parameters and / or the position of the target battery cell.
[0010] In some embodiments, controlling the light source of the first detection station to switch between multiple wavelengths of light includes: the light source of the first detection station switches between multiple wavelengths of light in a time-sharing stroboscopic manner; the wavelength of any light is greater than or equal to the wavelength of ultraviolet light and less than or equal to the wavelength of infrared light.
[0011] In some embodiments, controlling the light source of the first detection station to switch to a plurality of different wavelengths of light includes sequentially switching ultraviolet light, blue light, green light, orange light, red light, far infrared light, infrared light, and white light sources.
[0012] In some embodiments, the defect detection method for a battery cell explosion-proof valve further includes: determining the wavelength of each of a plurality of different wavelengths of light based on a type of defect to be detected.
[0013] In some embodiments, fusing multiple single-spectrum images of different wavelengths includes: fusing a single-spectrum image of an ultraviolet wavelength, a single-spectrum image of a blue light wavelength, a single-spectrum image of a green light wavelength, a single-spectrum image of an orange light wavelength, a single-spectrum image of a red light wavelength, and a single-spectrum image of a white light wavelength of the target battery cell; and / or fusing a single-spectrum image of an infrared wavelength and a single-spectrum image of a far-infrared wavelength of the target battery cell.
[0014] In some embodiments, determining a defect detection result of the explosion-proof valve of the target battery cell includes: determining a first defect detection result of the explosion-proof valve of the target battery cell based on the generated multispectral fusion image, the first defect at least indicating a planar defect of the explosion-proof valve of the target battery cell.
[0015] In some embodiments, determining a first defect detection result regarding the explosion-proof valve of a target battery cell includes: performing feature extraction on a generated multispectral fusion image based on a predetermined detection model to determine defective pixel information of an area where the explosion-proof valve of the target battery cell is located, the defective pixel information indicating defective pixel category information and corresponding defective pixel position information; determining a size of a defective area associated with the defective pixel category information based on the defective pixel category information and the corresponding defective pixel position information; and in response to determining that the size of the defective area is greater than a predetermined defective area size threshold, determining that the first defect detection result regarding the explosion-proof valve of the target battery cell indicates that there is a planar defect in the explosion-proof valve of the target battery cell.
[0016] In some embodiments, the planar defect includes at least one of the following: electrolyte splashing, corrosion, dirt, and foreign matter.
[0017] In some embodiments, fusing multiple single-spectrum images of different wavelengths includes: extracting any single-spectrum image in response to determining that a characteristic parameter of any single-spectrum image among the multiple single-spectrum images of different wavelengths meets a predetermined characteristic parameter threshold condition; and performing image fusion based on the characteristics of the extracted single-spectrum image to generate a multi-spectral fusion image of the target battery cell.
[0018] In some embodiments, the defect detection method for the battery cell explosion-proof valve also includes: acquiring three-dimensional point cloud data about the target battery cell; and calculating the height information of the top surface of the target battery cell based on the acquired three-dimensional point cloud data, so as to determine a second defect detection result of the explosion-proof valve of the target battery cell based on the height information, the second defect at least indicating a morphological defect of the explosion-proof valve of the target battery cell.
[0019] In some embodiments, calculating the height information of the top surface of the target battery cell includes: in response to a first defect detection result of the explosion-proof valve of the target battery cell indicating the presence of a predetermined defect type, calculating the height information of the top surface of the target battery cell based on the acquired three-dimensional point cloud data.
[0020] In some embodiments, determining a second defect detection result of the explosion-proof valve of the target battery cell based on height information includes: determining a mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image, so as to determine an area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data, the area of interest including at least a portion of the area where the explosion-proof valve is located; performing plane fitting on the determined area of interest, so as to calculate height information of the area of interest; and determining the second defect detection result of the explosion-proof valve of the target battery cell based on the calculated height information of the area of interest.
[0021] In some embodiments, determining the mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image includes: preprocessing the acquired three-dimensional point cloud data through bilateral filtering to remove abnormal data in the acquired three-dimensional point cloud data; extracting multiple first key points from the three-dimensional point cloud data after removing the abnormal data to determine the spatial conversion relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image; and determining the mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image based on the determined spatial conversion relationship.
[0022] In some embodiments, determining an area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data includes: determining the area of interest based on the first key point, the area of interest including at least an area where a first defect of the explosion-proof valve of the target battery cell is located.
[0023] In some embodiments, determining the area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data also includes: extracting multiple second key points in the area where the explosion-proof valve of the target battery cell is located, and / or determining multiple second key points among the first key points; determining the area where the first defect of the explosion-proof valve of the target battery cell is located based on the second key points; and determining the area where the first defect is located as the area of interest in the acquired three-dimensional point cloud data.
[0024] In some embodiments, the morphological defect includes at least one of the following: breakage and deformation.
[0025] According to a second aspect of the present disclosure, a computing device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; the memory storing instructions executable by the at least one processor, the instructions being executed by the at least one processor so as to enable the at least one processor to execute the method of the first aspect of the present disclosure.
[0026] In a third aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method of the first aspect of the present disclosure.
[0027] According to a fourth aspect of the present disclosure, a defect detection system for a battery cell explosion-proof valve is provided, comprising: a battery cell carrying device, configured to carry a target battery cell and having at least one inspection station; at least one light source, configured to respectively emit light of multiple different wavelengths to illuminate the target battery cell and emit laser to illuminate the target battery cell; at least one image acquisition device, configured to acquire single-spectrum images of multiple different wavelengths of the target battery cell and three-dimensional point cloud data of the target battery cell at at least one inspection station; and a computing device according to the second aspect of the present disclosure, configured to determine a defect detection result of the explosion-proof valve of the target battery cell based on the multiple single-spectrum images of different wavelengths acquired by the at least one image acquisition device.
[0028] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method according to the first aspect of the present disclosure when executed by a processor.
[0029] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference numerals are used throughout the drawings to represent the same components. In the drawings:
[0031] FIG1 shows a schematic diagram of a system for implementing a defect detection method for a battery cell explosion-proof valve according to an embodiment of the present invention.
[0032] FIG2 shows a schematic diagram of a top structure of a battery cell according to an embodiment of the present disclosure.
[0033] FIG3 shows a flow chart of a method for detecting defects in a battery cell explosion-proof valve according to an embodiment of the present disclosure.
[0034] FIG4 shows a flowchart of a method for determining a second defect detection result of an explosion-proof valve of a target battery cell according to an embodiment of the present disclosure.
[0035] FIG5 shows a flowchart of a method for generating a multispectral fusion image according to an embodiment of the present disclosure.
[0036] FIG6 shows a flowchart of a method for determining a first defect detection result according to an embodiment of the present disclosure.
[0037] FIG7 shows a flowchart of a method for determining a second defect detection result according to an embodiment of the present disclosure.
[0038] FIG8 shows a flowchart of a method for determining a mapping relationship according to an embodiment of the present disclosure.
[0039] FIG9 shows a flowchart of a method for determining a region of interest according to an embodiment of the present disclosure.
[0040] FIG10 is a schematic diagram showing imaging differences of lubricating oil under single spectra at different wavelengths according to an embodiment of the present disclosure.
[0041] FIG11 shows a schematic diagram of electrolyte sputtering image fusion according to an embodiment of the present disclosure.
[0042] FIG12 shows a schematic diagram of a plane fitting effect according to an embodiment of the present disclosure.
[0043] FIG13 shows a schematic structural diagram of a defect detection system for a battery cell explosion-proof valve according to an embodiment of the present disclosure.
[0044] FIG14 shows a schematic diagram of a second inspection station according to an embodiment of the present disclosure.
[0045] FIG15 shows a block diagram of an electronic device according to an embodiment of the present disclosure. Modes for Carrying Out the Invention
[0046] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.
[0047] As used herein, the term "including" and its variations represent open inclusion, i.e., "including but not limited to." Unless otherwise stated, the term "or" means "and / or." The term "based on" means "based at least in part on." The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment." The term "another embodiment" means "at least one additional embodiment." The terms "first," "second," etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.
[0048] With the widespread application of batteries in various fields, their safety issues are receiving increasing attention. As the core component of a battery, the battery cell performs core functions such as converting chemical energy into electrical energy, storing electrical energy, and charging and discharging. Therefore, battery cell safety testing is a crucial component of battery safety. The battery cell explosion-proof valve, as a safety device, can prevent fire or explosion caused by overheating or overvoltage. It automatically releases pressure when the internal pressure of the battery cell reaches a threshold, preventing excessive pressure and thus reducing the risk of explosion. Furthermore, beyond explosion-proofing, battery cell explosion-proof valves may also provide multiple functions, such as ventilation, waterproofing, dustproofing, and oil-proofing. Therefore, defect detection of battery cell explosion-proof valves is a crucial component of battery cell safety testing.
[0049] As described above, traditional battery cell explosion-proof valve inspection technologies, such as manual inspection or using a camera module to photograph the battery cell surface for image recognition to detect defects in the battery cell explosion-proof valve, rely heavily on the experience of professionals and use subjective judgment criteria. Different inspectors produce inconsistent and unstable test results, and manual inspection is inefficient and unable to adapt to the increasingly automated and intelligent production pace. Traditional inspection methods using camera modules and image recognition typically require capturing multiple sets of images, which is complex and computationally intensive, and cannot distinguish between defect types in the battery cell explosion-proof valve. In summary, the shortcomings of traditional battery cell explosion-proof valve defect detection methods are that they make it difficult to accurately and efficiently detect defects in battery cell explosion-proof valves.
[0050] In order to at least partially solve one or more of the above-mentioned problems and other potential problems, an exemplary embodiment of the present disclosure proposes a defect detection scheme for a battery cell explosion-proof valve. In the scheme of the present disclosure, by acquiring multiple single-spectrum images of different wavelengths about the target battery cell (each single-spectrum image at least indicates the explosion-proof valve area of the target battery cell); and fusing multiple single-spectrum images of different wavelengths to generate a multi-spectral fusion image about the target battery cell, the present disclosure can more comprehensively acquire the imaging details of the battery cell explosion-proof valve area in different single-spectrum images based on the difference in imaging effects of the same object under different wavelength spectra, and the multi-spectral fusion image generated based on this can have more obvious features about the explosion-proof valve defect; then, based on the generated multi-spectral fusion image, determine the defect detection result of the explosion-proof valve of the target battery cell. Therefore, the present disclosure can accurately and efficiently obtain the defect detection result of the explosion-proof valve of the target battery cell.
[0051] Figure 1 shows a schematic diagram of a system 100 for implementing a defect detection method for battery cell explosion-proof valves according to an embodiment of the present invention. As shown in Figure 1 , system 100 includes a computing device 110, at least one light source (e.g., light source 140), at least one image acquisition device (e.g., image acquisition device 120), a target battery cell 200, and a battery cell support device (not shown). Computing device 110, light source 140, and image acquisition device 120 can exchange data via a network (e.g., the Internet, a communications network, etc.).
[0052] As for the image acquisition device 120 , it is configured at at least one inspection station on the battery cell supporting device to acquire single-spectrum images of a plurality of different wavelengths of the target battery cell 200 .
[0053] Regarding the battery cell carrying device, it is used to carry the target battery cell 200. The battery cell carrying device has at least one inspection station, for example, a first inspection station and a second inspection station.
[0054] The light source 140 is configured to emit at least a plurality of lights of different wavelengths for illuminating target cells.
[0055] It should be understood that the configuration of the battery cell carrier, the parameters of the light source, the type and parameters of the image acquisition device, and the relative positions of these devices can all be adjusted based on the image acquisition requirements of the target battery cell to meet the image acquisition requirements of the target battery cell in this solution.
[0056] The computing device 110 is configured to determine a defect detection result for the explosion-proof valve of a target battery cell based on multiple single-spectral images at different wavelengths captured by at least one image acquisition device. Specifically, the computing device 110 is configured to, for example, acquire multiple single-spectral images at different wavelengths of the target battery cell, each single-spectral image indicating at least the explosion-proof valve region of the target battery cell; fuse the multiple single-spectral images at different wavelengths to generate a multispectral fused image of the target battery cell; and determine a defect detection result for the explosion-proof valve of the target battery cell based on the generated multispectral fused image.
[0057] The computing device 110 may include one or more processing units, including specialized processing units such as GPUs, FPGAs, and ASICs, as well as general-purpose processing units such as CPUs. Furthermore, one or more virtual machines may also be running on each computing device 110. In some embodiments, the computing device 110 and the battery cell explosion-proof valve defect detection system 130 may be integrated or separately configured. In some embodiments, the computing device 110 may include, for example, a single-spectrum image acquisition module 112, a multi-spectral fusion image generation module 114, and a defect detection result generation module 116.
[0058] The single spectrum image acquisition module 112 is configured to acquire a plurality of single spectrum images of different wavelengths of the target battery cell, each single spectrum image at least indicating an explosion-proof valve region of the target battery cell.
[0059] The multi-spectral fusion image generating module 114 is configured to fuse a plurality of single-spectrum images of different wavelengths to generate a multi-spectral fusion image of the target battery cell.
[0060] The defect detection result generating module 116 is configured to determine a defect detection result of the explosion-proof valve of the target battery cell based on the generated multi-spectral fusion image.
[0061] Figure 2 shows a schematic diagram of the top structure of a battery cell according to an embodiment of the present disclosure. Figure 2 shows a schematic diagram of the structure of a battery cell top patch 201, and the battery cell top patch 201 includes an explosion-proof valve 202, a battery cell QR code 203, a positive electrode column 204 and a negative electrode column 205; wherein the explosion-proof valve 202 includes an explosion-proof valve edge 2021 (the black edge portion of 202 in the figure) and a PP film (PP SHEET FOR OPP, polypropylene film) 2022 (the portion other than the black edge of 202).
[0062] The battery cell QR code 203 is a special QR code used to track the manufacturing and transportation process of the battery and provide information about the battery for authentication. The battery cell QR code usually contains information such as the battery's serial number, batch number, manufacturing date, manufacturer, and battery specifications. In the solution provided by the present disclosure, the target battery cell is identified by the battery cell QR code 203, and the image data collected by the target battery cell at each inspection station is associated, and the defect detection results of the target battery cell are associated with its corresponding battery cell QR code, so as to record, query and trace the detection results of the target battery cell explosion-proof valve, and facilitate the timely discovery and resolution of potential quality problems and improve product quality.
[0063] Figure 3 shows a flow chart of a method 300 for defect detection of a battery cell explosion-proof valve according to an embodiment of the present disclosure. Figure 10 shows a schematic diagram of the differences in imaging of lubricating oil under single spectra at different wavelengths according to an embodiment of the present disclosure. Method 300 is further described below in conjunction with Figures 2, 3, and 10. Method 300 can be executed by the computing device 110 shown in Figure 1 or by the electronic device 1500 shown in Figure 15. It should be understood that method 300 may also include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0064] In step 302 , the computing device 110 obtains a plurality of single-spectrum images of different wavelengths of a target battery cell, where each single-spectrum image at least indicates an explosion-proof valve region of the target battery cell.
[0065] Regarding multiple different wavelengths, it refers to different wavelength ranges, that is, multiple single-spectrum images of different wavelengths include at least two single-spectrum images, and are images of the target battery cell under light in two different wavelength ranges respectively; regarding light in different wavelength ranges, it includes at least two of the following: ultraviolet light, blue light, green light, orange light, red light, far infrared light, infrared light and white light.
[0066] Regarding single-spectrum images, each single-spectrum image is an image collected under a specific wavelength of light. The specific wavelength range can be a fixed wavelength or a wavelength range. It should be noted that the multiple different wavelengths of light used to collect single-spectrum images in the embodiments of the present disclosure can be a monochromatic light or a polychromatic light, and can include both visible light and invisible light (such as infrared light and ultraviolet light). Since light of different wavelengths is reflected and absorbed differently on the surface of an object, the same object will have different imaging differences under different wavelength single spectra, for example, different brightness and contrast will appear during imaging. In some embodiments, the characteristic parameters of the single-spectrum image are related to one or more of the following: texture, grayscale, edge, and contour.
[0067] Please refer to Figure 10, which illustrates the imaging differences of lubricating oil under different spectra, where 10a) illustrates the imaging effect of lubricating oil under 370nm ultraviolet light, 10b) illustrates the imaging effect of lubricating oil under 600nm visible light, and 10c) illustrates the imaging effect of lubricating oil under 900nm infrared light. From the three imaging effects illustrated in Figure 10, it can be found that the imaging effect of each image is different (such as contrast, grayscale, brightness, texture, edges and details, etc.), that is, each single-spectrum image captures different features for the same object.
[0068] Please refer to the battery cell top patch 201 and the explosion-proof valve area 202 of the battery cell shown in FIG2 . Each single spectrum image acquired in the embodiment of the present disclosure includes at least the explosion-proof valve area of the target battery cell.
[0069] At step 304 , the computing device 110 fuses a plurality of single-spectrum images at different wavelengths to generate a multi-spectral fused image of the target battery cell.
[0070] Regarding multispectral fusion images, single-spectrum images provide less information, and some defects have poor imaging quality under certain single-spectrum conditions, making them difficult to identify defect features. The solution provided by the disclosed embodiments uses multiple wavelengths of light to perform single-spectrum imaging of the target cell and fuses the imaging results of the different wavelengths to generate a multispectral fusion image with richer information. As a result, the multispectral fusion image of the target cell can provide more information than a single-spectrum image, thereby improving image quality and the ability to identify explosion-proof valve defects.
[0071] At step 306 , the computing device 110 determines a defect detection result of the explosion-proof valve of the target battery cell based on the generated multispectral fusion image.
[0072] Regarding the defect detection results, in some embodiments, it can indicate whether the explosion-proof valve of the target battery cell has defects, and can also be used to indicate the type of defects in the explosion-proof valve of the target battery cell. Regarding the defect types of the explosion-proof valve, for example, electrolyte sputtering, primary corrosion, secondary corrosion, dirt, primary foreign matter, secondary foreign matter, breakage and deformation, etc.; among them, the primary and secondary levels are of different degrees, such as the primary level is lighter than the secondary level, and the judgment threshold is determined according to the actual production line standard.
[0073] In some embodiments, the defect detection result includes a first defect detection result and a second defect detection result. The first defect at least indicates a planar defect of the explosion-proof valve of the target battery cell. Planar defects include, for example, electrolyte sputtering, primary corrosion, secondary corrosion, dirt, primary foreign matter, and secondary foreign matter. Both primary corrosion and secondary corrosion may be surface corrosion caused by electrolyte sputtering, but the degree of corrosion is different. The specific standards for primary corrosion and secondary corrosion may be adjusted according to the needs of different production lines. Primary foreign matter may be sweat stains that have little effect on the explosion-proof valve. The specific standards for secondary foreign matter and primary foreign matter may also be adjusted according to the needs of different production lines. The second defect at least indicates a morphological defect of the explosion-proof valve of the target battery cell.
[0074] In the above scheme, by acquiring multiple single-spectrum images of different wavelengths of the target battery cell, and then fusing the multiple single-spectrum images of different wavelengths to generate a multi-spectral fusion image of the target battery cell, the present disclosure can be based on the difference in imaging effects of the same object under different wavelength spectra. The generated multi-spectral fusion image can provide more information than the single-spectrum image, thereby improving the image quality and making the defect characteristics of the explosion-proof valve more obvious. The generated multi-spectral fusion image is then detected through a predetermined detection model to determine the defect detection result of the explosion-proof valve of the target battery cell. Therefore, the present disclosure can accurately and efficiently obtain the defect detection result of the explosion-proof valve of the target battery cell.
[0075] In some embodiments, a method for acquiring multiple single-spectrum images of a target battery cell at different wavelengths includes: controlling a light source of a first inspection station to switch between multiple wavelengths of light during a single exposure time, so that a first image acquisition device can acquire multiple single-spectrum images of the target battery cell at different wavelengths, wherein the station parameters of the first inspection station are related to the parameters of the first image acquisition device, the light source parameters, and / or the position of the target battery cell. Referring to FIG. 13 , which illustrates a defect detection system 130 for a battery cell explosion-proof valve, the target battery cell 200 is located at the first inspection station. A detailed description of the first inspection station is provided below in the description of system 130 and is not repeated here.
[0076] Regarding the parameters of the first image acquisition device, they include, for example, at least the type, shooting angle, shooting parameters, and device position of the first image acquisition device.
[0077] Regarding light source parameters, they include, for example, at least the light source position, light source layout, light source type, light source wavelength, and light source spectrum switching rules.
[0078] Regarding the position of the target battery cell, when the first image acquisition device is performing image acquisition on the target battery cell at the first inspection station, the target battery cell and the first image acquisition device are relatively stationary.
[0079] Therefore, within a single exposure time, the light source of the first inspection station can switch to light of different wavelengths to cooperate with the first image acquisition device to acquire multiple single-spectrum images of different wavelengths about the target battery cell, and can acquire multiple single-spectrum images of different wavelengths of multiple target battery cells at one time, which is convenient to operate and greatly saves the time required for image acquisition; and the relative position of the first target battery cell and the first image acquisition device remains unchanged during image acquisition, so that the shooting content and orientation of the multiple single-spectrum images acquired by the first image acquisition device are consistent, and no additional image calibration or alignment operations are required to make the position, size, proportion, etc. of the explosion-proof valve in the multiple single-spectrum images consistent, which can reduce the amount of image processing calculations and save image processing time, so as to improve the efficiency of multi-spectral fusion, thereby improving the detection efficiency and speeding up the production rhythm.
[0080] In some embodiments, controlling the light source of the first detection station to switch between multiple wavelengths of light includes: the light source of the first detection station switches between multiple wavelengths of light in a time-sharing stroboscopic manner; the wavelength of any light is greater than or equal to the wavelength of ultraviolet light and less than or equal to the wavelength of infrared light.
[0081] Regarding the wavelength of any light, it is a fixed wavelength or a wavelength range; for example, it is a wavelength or wavelength range of monochromatic light, a wavelength range of polychromatic light, a wavelength or wavelength range of infrared light, or a wavelength or wavelength range of ultraviolet light. It should be understood that the optional range of the wavelength of any light in the embodiment of the present disclosure includes any part of the wavelength range included in ultraviolet light, visible light, and infrared light; since the wavelength range of ultraviolet light is 10 nanometers to 400 nanometers, the wavelength range of visible light is about 380 nanometers to about 780 nanometers, and the wavelength range of infrared light is 750 nanometers to 1000 microns; therefore, the wavelength of any light is a part of the wavelength range of 10 nanometers to 1000 microns, which can be a specific fixed wavelength, a continuous wavelength range, or a combination of multiple discontinuous wavelength ranges to form a kind of light.
[0082] Therefore, the embodiment of the present disclosure performs light source switching and image acquisition at the first inspection station in a time-sharing stroboscopic manner, which can make the light source switching and shutter frequency more synchronized, so that the imaging quality is not affected during the light source switching, so as to improve the image quality while increasing the image acquisition speed, obtain higher resolution images and higher contrast single spectrum images, so as to improve the battery cell defect detection effect and detection efficiency.
[0083] In some embodiments, controlling the light source of the first detection station to switch to a plurality of different wavelengths of light includes sequentially switching ultraviolet light, blue light, green light, orange light, red light, far infrared light, infrared light, and white light sources.
[0084] In some embodiments, the method further includes determining a wavelength of each of the plurality of different wavelengths of light based on a type of defect to be detected.
[0085] In this way, the number of images taken during image acquisition can be reduced, the detection rhythm can be improved, and light with better sensitivity can be selected for specific detection items to achieve rapid detection.
[0086] In some embodiments, fusing multiple single-spectrum images of different wavelengths includes: fusing a single-spectrum image of an ultraviolet wavelength, a single-spectrum image of a blue light wavelength, a single-spectrum image of a green light wavelength, a single-spectrum image of an orange light wavelength, a single-spectrum image of a red light wavelength, and a single-spectrum image of a white light wavelength of the target battery cell; and / or fusing a single-spectrum image of an infrared wavelength and a single-spectrum image of a far-infrared wavelength of the target battery cell.
[0087] Regarding the method of fusing multiple single-spectrum images of different wavelengths, in some embodiments, it includes: fusing a single-spectrum image of a target battery cell at an infrared wavelength and a single-spectrum image at a far-infrared wavelength to detect the presence of foreign matter beneath the PP film of the explosion-proof valve of the target battery cell. It should be understood that the edges of the PP film for OPP explosion-proof valves (PP sheet for OPP, polypropylene film) are opaque, making it difficult for conventional image detection solutions to detect foreign matter on the aluminum foil of the explosion-proof valve. However, infrared and far-infrared wavelengths have a certain degree of penetration, which can detect foreign matter on the surface of the aluminum foil directly below the PP film for OPP explosion-proof valves (PP sheet for OPP, polypropylene film). The present disclosure can generate a multispectral fusion image by selecting a single-spectral image of an infrared wavelength and a single-spectral image of a far-infrared wavelength of a target battery cell from the collected single-spectral images. Then, based on the generated multispectral fusion image, a predetermined detection model is used to determine whether there is foreign matter under the PP film (PP SHEET FOR OPP, polypropylene film) of the explosion-proof valve of the target battery cell, and the specific type of foreign matter defect, such as primary foreign matter or secondary foreign matter, as well as the possible types of foreign matter. Among them, the primary foreign matter and the secondary foreign matter only differ in the degree of foreign matter, such as the primary degree is lighter than the secondary degree, and the judgment threshold is determined according to the actual production line standard.
[0088] As a result, two different inspection items can be performed during the first defect detection, such as detecting electrolyte, corrosion and foreign matter; or only the single spectrum image of the infrared wavelength and the single spectrum image of the far-infrared wavelength of the target battery cell can be integrated to detect foreign matter.
[0089] In some embodiments, determining a defect detection result of the explosion-proof valve of the target battery cell includes: determining a first defect detection result of the explosion-proof valve of the target battery cell based on the generated multispectral fusion image, the first defect at least indicating a planar defect of the explosion-proof valve of the target battery cell.
[0090] FIG5 illustrates a flow chart of a method 500 for generating a multispectral fusion image according to an embodiment of the present disclosure. FIG11 illustrates a schematic diagram of electrolyte sputtering image fusion according to an embodiment of the present disclosure. Method 500 is further described below in conjunction with FIG5 and FIG11. Method 500 may be executed by computing device 110 as shown in FIG1 or by electronic device 1500 as shown in FIG15. It should be understood that method 500 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0091] In step 502 , if it is determined that the characteristic parameter of any single-spectrum image among the multiple single-spectrum images of different wavelengths meets a predetermined characteristic parameter threshold, the computing device 110 extracts any single-spectrum image.
[0092] Therefore, it is not necessary to select all single-spectrum images of the target battery cell for multi-spectral fusion. It is only necessary to select images with more obvious features to exclude single-spectrum images generated by spectra that are insensitive to certain defect features, thereby saving the amount of image processing computation and improving detection efficiency.
[0093] At step 504 , the computing device 110 performs image fusion based on the extracted features of the single-spectrum image to generate a multi-spectral fusion image of the target battery cell.
[0094] Regarding image fusion, it at least includes pixel fusion through operations such as weighted averaging, maximization, minimization, etc., for example, various transformation fusions such as Fourier transform, wavelet transform, etc. are performed on the image, and for example, images are fused based on energy functions. Through the image fusion operation, the features of the explosion-proof valve defects in the generated multispectral fusion image can be more easily identified.
[0095] For example, the following formula (1) illustrates the algorithm for generating a multi-spectral fusion image of a target cell: F(i,j)=w1A(i,j)+w2B(i,j) (1)
[0096] In formula (1), A represents a single-spectrum image to be fused. B represents another single-spectrum image to be fused. F represents the multispectral fused image obtained by fusion. (i, j) represents the position of the defective pixel in the image. w1 and w2 represent weight information, which can be adjusted according to the fusion effect.
[0097] In some embodiments, before generating the multi-spectral fusion image of the target battery cell, the method includes: performing an image enhancement operation on the extracted single-spectrum image to improve feature recognition of the single-spectrum image.
[0098] Regarding the method of image enhancement, it at least includes: for example, adjusting the histogram of the pixel values of the local or overall image to adjust the image brightness distribution; for example, reducing noise and enhancing the edge features of the image through median filtering, Gaussian filtering, etc.; for example, enhancing the frequency domain information by performing frequency domain transformation on the image; for example, stretching, transforming, equalizing, adjusting the contrast of the image locally or globally to improve the recognizability of the defect features of the explosion-proof valve in the image.
[0099] Please refer to Figure 11. Image 11a and Image 11b in Figure 11 respectively illustrate the imaging effects of at least the explosion-proof valve area of the target battery cell under two different single spectra. Images 11a and 11b indicate that there is a shadow on the lower left side of the battery cell explosion-proof valve, which is suspected to be caused by electrolyte sputtering. By fusing images 11a and 11b to generate a multi-spectral fusion image 11c of the target battery cell, it can be clearly seen that the shadow in the lower left corner of image 11c is more obvious. By performing feature extraction and other operations on the multi-spectral fusion image 11c, the type and position of the defective pixel can be more easily identified and confirmed. Therefore, the method provided by the embodiment of the present disclosure can obtain defect detection results more efficiently and accurately through multi-spectral fusion images.
[0100] FIG6 illustrates a flow chart of a method 600 for determining a first defect detection result according to an embodiment of the present disclosure. Method 600 may be executed by computing device 110 as shown in FIG1 , or may be executed by electronic device 1500 as shown in FIG15 . It should be understood that method 600 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0101] In step 602 , the computing device 110 performs feature extraction on the generated multispectral fusion image based on a predetermined detection model to determine defective pixel information in the area where the explosion-proof valve of the target battery cell is located, where the defective pixel information indicates defective pixel category information and corresponding defective pixel location information.
[0102] Regarding the predetermined detection model, it can be constructed and trained based on models such as deep learning models, convolutional neural networks and semantic segmentation models, for example, based on network models such as FCN, U-Net, SegNet, DeepLab V3+, PSPNet, transformer, VGG, ResNet, Inception, MobileNet, EfficientNet, Faster R-CNN, YOLO, SSD, Mask R-CNN and RetinaNet.
[0103] As sample data for training a predetermined detection model, a large number of single spectra and fused spectrum images of target battery cells, including at least the various defects of their explosion-proof valves, can be collected. Labeling is then performed on the collected images associated with the various defects in the explosion-proof valves, indicating the defect pixel locations, corresponding defect types, and target battery cell models. The labeled images are then used as sample data for training the predetermined detection model.
[0104] Regarding defective pixel categories, the defect categories can be set based on the detection requirements of the production line. For example, for planar defects, there are multiple categories including electrolyte sputtering, corrosion (such as primary corrosion, secondary corrosion), dirt, and foreign matter (such as primary foreign matter, secondary foreign matter).
[0105] The defective pixel position information can indicate the position of the defective pixel in the generated multispectral fusion image. For example, in the multispectral fusion image t, the spatial position of the defective pixel q in the multispectral fusion image t is represented by q=(tx, ty).
[0106] At step 604 , the computing device 110 determines a size of a defective region associated with the defective pixel category information based on the defective pixel category information and the corresponding defective pixel location information.
[0107] For example, there are n pixels whose defective pixel category is “dirty”, and the size occupied by these n “dirty” pixels in the multispectral fusion image t is Sn.
[0108] At step 606 , if it is determined that the size of the defective region is greater than the predetermined defective region size threshold, the computing device 110 determines that the first defect detection result regarding the explosion-proof valve of the target battery cell indicates that the explosion-proof valve of the target battery cell has a planar defect.
[0109] Regarding the predetermined defect area size threshold, the defect size thresholds corresponding to different defective pixel categories may be different. It should be understood that different defects have different impacts on the safety of the battery cell, so their size thresholds are obviously different. For example, electrolyte sputtering has less impact on battery cell safety, and its corresponding predetermined defect size threshold may be greater than the predetermined defect size threshold corresponding to the dirt defect, the latter of which has a greater impact on battery cell safety.
[0110] As for the predetermined defect area size threshold, it can be adjusted according to the quality inspection requirements of the production line and different battery products. Therefore, the size threshold corresponding to different defect pixel categories is adjustable.
[0111] Regarding the presence of planar defects, the solution provided by the embodiments of the present disclosure can not only determine whether the explosion-proof valve of the target battery cell has a planar defect, but also determine the type of the planar defect, thereby obtaining a more accurate explosion-proof valve defect detection result.
[0112] Figure 4 shows a flow chart of a method 400 for determining a second defect detection result for an explosion-proof valve of a target battery cell according to an embodiment of the present disclosure. Method 400 can be executed by computing device 110 as shown in Figure 1 , or by electronic device 1500 as shown in Figure 15 . It should be understood that method 400 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0113] In step 402 , the computing device 110 obtains three-dimensional point cloud data about a target battery cell.
[0114] In step 404, the computing device 110 calculates height information of the top surface of the target battery cell based on the acquired three-dimensional point cloud data, so as to determine a second defect detection result of the explosion-proof valve of the target battery cell based on the height information, where the second defect at least indicates a morphological defect of the explosion-proof valve of the target battery cell.
[0115] Regarding the height information of the top surface of the target battery cell, it should be understood that the explosion-proof valve of the target battery cell in the embodiment of the present disclosure is set on the top surface of the battery cell, such as the battery cell top patch 201 shown in Figure 2. The height of the top surface of the target battery cell can be calculated through the three-dimensional point cloud data of the target battery cell. The calculated height information can be used to determine whether there is an abnormality in the explosion-proof valve. For example, if there is a large difference in the height information of the edge area 202 of the explosion-proof valve, discontinuous height information, or a large difference in the height of the PP film (PP SHEET FOR OPP, polypropylene film) area, it may be that there is a morphological defect, such as breakage, deformation, etc.
[0116] In some embodiments, if the first defect detection result of the explosion-proof valve of the target battery cell indicates the presence of a predetermined defect type, the computing device 110 calculates height information of the top surface of the target battery cell based on the acquired three-dimensional point cloud data.
[0117] For example, if the predetermined defect types are secondary corrosion, contamination, and secondary foreign matter, then if the first defect detection result of the explosion-proof valve of the target battery cell indicates any of secondary corrosion, contamination, and secondary foreign matter, then the height information of the top surface of the target battery cell needs to be calculated. In this case, if the first defect detection result of the explosion-proof valve of the target battery cell does not indicate secondary corrosion, contamination, or secondary foreign matter, for example, but only indicates no defect, electrolyte splashing, primary corrosion, and / or primary foreign matter, then the height information of the top surface of the target battery cell does not need to be calculated.
[0118] It should be noted that some planar defects are considered to have no possibility of morphological defects in the production line standards. For these target battery cells with planar defects, there is no need to perform height calculation after obtaining their three-dimensional point cloud data, and the default second defect detection result is no defect. In this case, it is only necessary to store the three-dimensional point cloud data of the target battery cells so that the second defect detection (such as random inspection) of these battery cells can be carried out through the stored three-dimensional point cloud data and multi-spectral fusion images during the quality inspection process.
[0119] Therefore, by excluding target cells that are considered to have no morphological defects in the production line standards from the height information calculation, the amount of calculation can be significantly saved and the detection efficiency can be improved.
[0120] In some embodiments, three-dimensional point cloud data about the target cell is obtained from a second inspection station. At the second inspection station, a laser generator serving as a light source and a second image capture device are positioned above the top surface of the target cell. The laser generator and the second image capture device are positioned at a predetermined angle so that the second image capture device receives light reflected from the top surface of the target cell. Referring to FIG. 14 , which illustrates target cell 200 at the second inspection station, a detailed description of the second inspection station is provided below in the description of system 130 and is not further elaborated here.
[0121] It should be understood that based on the reflection of light and the principle of similar triangles, when the top surface of the target battery cell, such as the area where the explosion-proof valve is located, changes in height due to morphological defects, the angle of light after reflection will change, causing the pixels in the point cloud data collected by the second image acquisition device to shift. When the pixel movement range exceeds the predetermined threshold, it means that this area may have defects such as deformation and damage.
[0122] Figure 7 illustrates a flow chart of a method 700 for determining a second defect detection result according to an embodiment of the present disclosure. Figure 12 illustrates a schematic diagram of a plane fitting effect according to an embodiment of the present disclosure. Method 700 is further described below in conjunction with Figures 7 and 12. Method 700 may be executed by computing device 110 as shown in Figure 1 or by electronic device 1500 as shown in Figure 15. It should be understood that method 700 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0123] In step 702, the computing device 110 determines a mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image to determine a region of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data, where the region of interest includes at least a portion of the area where the explosion-proof valve is located.
[0124] In step 704 , the computing device 110 performs plane fitting on the determined region of interest to calculate height information of the region of interest.
[0125] In some embodiments, if the first defect detection result of the explosion-proof valve of the target battery cell does not meet the predetermined defect type, after determining the area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data, there is no need to calculate the height information of the area of interest. It is only necessary to associate the area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data with the multispectral fusion image and the QR code of the battery cell and store them in the detection database so as to perform random inspection or traceability on the target battery cell.
[0126] Please refer to Figure 12. Image 12a in Figure 12 shows a multispectral fusion image of a target battery cell. A dark shadow can be seen in the upper left area of the explosion-proof valve in Image 12a. In the absence of three-dimensional height information in a multispectral fusion image, it is often difficult to determine whether the defect type is dirt, corrosion, deformation, or damage. Image 12b in Figure 12 shows a schematic diagram of the plane fitting effect of the target battery cell. It uses an intuitive three-dimensional image rendering method to display the height information of the explosion-proof valve area in Image 12a. Combined with Image 12b, the second detection result of the target battery cell can be determined as the presence of a morphological defect, namely, damage. Therefore, through the combined judgment of planar defect information and height information of the area of interest, this solution can accurately confirm and distinguish multiple defect types in the target battery cell.
[0127] At step 706 , the computing device 110 determines whether the second defect detection result of the explosion-proof valve of the target battery cell indicates that the explosion-proof valve of the target battery cell has a morphological defect based on the calculated height information of the region of interest.
[0128] Height information can be used to determine whether the explosion-proof valve of the target battery cell has morphological defects, as well as the type of morphological defects. For example, if there are differences in the height information of the region of interest but the pixels are continuous, it may be that the explosion-proof valve has deformed. For example, if the height information of the region of interest shows a sudden change, such as a sudden change in pixel height and poor pixel continuity, it may be damaged. It should be understood that the morphological defect type corresponding to specific height information can be determined by using the change characteristics of height information collected previously for various morphological defects.
[0129] Therefore, through the mapping relationship between the acquired three-dimensional point cloud data and the multispectral fusion image, the position of the explosion-proof valve in the multispectral fusion image can be corresponded to the area where the explosion-proof valve is located in the three-dimensional point cloud data, thereby facilitating the determination of the area of interest and the area where height information calculation is required.
[0130] It should be understood that in order to save computing power and improve detection efficiency, determining that the area of interest includes at least a part of the area where the explosion-proof valve is located can effectively reduce the area of height calculation and improve the efficiency of plane fitting. When performing plane fitting, the areas of non-interest can be quickly rendered, and more specific height information can be calculated for the areas of interest. In this way, a three-dimensional rendering of the top surface of the target battery cell can be quickly generated, thereby improving detection efficiency without missing the height information of the defect location, thereby speeding up the production pace.
[0131] FIG8 shows a flowchart of a method 800 for determining a mapping relationship according to an embodiment of the present disclosure. The method 800 may be executed by the computing device 110 shown in FIG1 , or may be executed at the electronic device 1500 shown in FIG15 . It should be understood that the method 800 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0132] At step 802 , the computing device 110 pre-processes the acquired three-dimensional point cloud data via bilateral filtering to remove abnormal data in the acquired three-dimensional point cloud data.
[0133] It should be understood that the use of bilateral filtering can effectively remove noise in point cloud data while retaining the details and edge information in the point cloud data, and can also smooth the point cloud data to make the point cloud data smoother and more continuous; it can improve the quality and usability of three-dimensional point cloud data.
[0134] In step 804 , the computing device 110 extracts a plurality of first key points from the three-dimensional point cloud data after removing abnormal data, so as to determine a spatial conversion relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image.
[0135] The first key points, for example, are pixel locations that can characterize the spatial positional characteristics of the top surface of the battery cell, thereby determining the spatial transformation relationship between the three-dimensional point cloud data and the generated multispectral fusion image. For example, referring to Figure 2 , at least two vertices of the battery cell's QR code 203 can be used as first key points. For example, pixels along the edge of a pole can be used as first key points, as can pixels along the edge of the explosion-proof valve 2021. It should be understood that the predetermined key point selection rules can be used to select multiple corresponding first key points from the generated multispectral fusion image and the acquired three-dimensional point cloud data to quickly determine the spatial transformation relationship between the two.
[0136] In step 806 , the computing device 110 determines a mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image based on the determined spatial conversion relationship.
[0137] In some embodiments, for example, the intrinsic parameters and extrinsic parameters of a first image acquisition device for acquiring a single-spectrum image and a second image acquisition device for acquiring three-dimensional point cloud data are respectively determined through camera calibration, wherein the intrinsic parameters include internal parameters of the camera such as focal length and optical center, and the extrinsic parameters include the position (rotation and translation) of the camera in the world coordinate system; posture estimation is performed based on the intrinsic and extrinsic parameters of the two image acquisition devices, and key points a and b in the multispectral fusion image are matched with corresponding points a' and b' in the three-dimensional point cloud data, and the relative rotation and translation transformation matrix of the multispectral fusion image is calculated, thereby determining the spatial transformation relationship matrix between the two; thereby achieving the mapping of the generated multispectral fusion image to the coordinate system of the acquired three-dimensional point cloud data, so as to determine the area of interest.
[0138] In some embodiments, for example, the optimal mapping function between the multispectral fusion image and the three-dimensional point cloud data may be found by using the least squares method.
[0139] In some embodiments, determining an area of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data includes: determining the area of interest based on the first key point, the area of interest including at least an area where a first defect of the explosion-proof valve of the target battery cell is located.
[0140] Therefore, the location of the area of interest can be determined through the first key point. By extracting the key point once, it is possible to simultaneously determine the mapping relationship between the three-dimensional point cloud data and the multispectral fusion image, and at least determine that the area where the first defect of the explosion-proof valve of the target battery cell is located is the area of interest.
[0141] FIG9 shows a flowchart of a method 900 for determining a region of interest according to an embodiment of the present disclosure. The method 900 may be executed by the computing device 110 shown in FIG1 , or may be executed at the electronic device 1500 shown in FIG15 . It should be understood that the method 900 may include additional steps not shown and / or may omit steps shown, and the scope of the present disclosure is not limited in this respect.
[0142] In step 902 , the computing device 110 extracts a plurality of second key points in the area where the explosion-proof valve of the target battery cell is located, and / or determines a plurality of second key points from the first key points.
[0143] In step 904 , a region where a first defect of the explosion-proof valve of the target battery cell is located is determined based on the second key point.
[0144] At step 906 , the computing device 110 determines the area where the first defect is located as a region of interest in the acquired three-dimensional point cloud data.
[0145] Regarding the second key point, it is, for example, certain key points of the area where the first defect is located, such as pixels about the edge information of the area where the first defect is located. For example, through the mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image, the area where the first defect pixel is located determined in the multispectral fusion image is projected into the corresponding area in the three-dimensional point cloud data to determine the area where the first defect is located in the three-dimensional point cloud data.
[0146] In some embodiments, the first key points include key points of the area where the first defect is located, and the key points of the area where the first defect is located are determined as second key points from the first key points; when the first key points do not include key points of the area where the first defect is located, or the key points of the area where the first defect is located in the first key points are insufficient to determine the area of interest (for example, the number and / or position are insufficient, making it impossible to determine the area of interest), the second key points are determined by extracting key points again to determine the area of interest.
[0147] Therefore, through the second key point, the position of the first defect of the explosion-proof valve of the target battery cell in the three-dimensional point cloud data can be determined, and the area of interest can be narrowed to the area near the first defect, so as to further reduce the amount of calculation during height calculation and improve detection efficiency.
[0148] FIG13 shows a schematic structural diagram of a defect detection system 130 for a battery cell explosion-proof valve according to an embodiment of the present disclosure.
[0149] As shown in FIG13 , a defect detection system 130 for a battery cell explosion-proof valve includes a battery cell supporting device (such as 132 ), one or more light sources (such as 134 ), and one or more image acquisition devices (such as 136 ).
[0150] Wherein, the battery cell carrying device is configured to carry the target battery cell and has at least one inspection station;
[0151] at least one light source configured to respectively emit light of a plurality of different wavelengths for illuminating a target battery cell and emit laser light for illuminating the target battery cell;
[0152] At least one image acquisition device is configured to acquire single-spectrum images of a plurality of different wavelengths of the target battery cell and three-dimensional point cloud data of the target battery cell at the at least one inspection station;
[0153] The battery cell explosion-proof valve defect detection system 130 is configured, according to the computing device disclosed in the embodiment of the present disclosure, to determine a defect detection result of the explosion-proof valve of the target battery cell based on multiple single-spectrum images of different wavelengths captured by at least one image acquisition device.
[0154] The target cell 200 is carried on the cell carrying device 132 shown in FIG13 . At this time, the target cell 200 is in the first detection station. The first detection station can at least be used to collect single-spectrum images of multiple different wavelengths about the target cell 200 .
[0155] The light source 134 shown in FIG. 13 is configured to emit light of a plurality of different wavelengths for respectively illuminating target cells.
[0156] In some embodiments, the light source 134 includes a plurality of LED sub-light sources that are evenly distributed in a ring shape.
[0157] In some embodiments, the light source 134 is further configured to switch between at least two different wavelengths of light based on instructions from the computing device.
[0158] In some embodiments, the at least two different wavelengths of light include two or more of the following: ultraviolet light, blue light, green light, orange light, red light, far infrared light, infrared light, and white light.
[0159] The image acquisition device 136 is configured to acquire single-spectrum images of a plurality of different wavelengths of the target battery cell at at least one inspection station.
[0160] In some embodiments, the light source further includes a laser generator configured to emit laser light to illuminate the target battery cell.
[0161] FIG14 shows a schematic diagram of a second inspection station according to an embodiment of the present disclosure. Referring to FIG14 , the system 130 further includes a second cell carrier 152 , a second light source 154 , and a second image acquisition device 156 .
[0162] In some embodiments, the battery cell supporting device includes a first battery cell supporting device 132 and a second battery cell supporting device 152 .
[0163] In some embodiments, the image acquisition device includes a first image acquisition device 136 and a second image acquisition device 156 .
[0164] In some embodiments, the image acquisition device 136 and the second image acquisition device 156 can be integrated into the same image acquisition device as image acquisition modules with different functions.
[0165] In some embodiments, the light source 134 and the second light source 154 can be integrated into the same light source device as different light-emitting modules, providing different lights when collecting single-spectrum images and collecting three-dimensional point cloud images, respectively.
[0166] In some embodiments, the second battery cell supporting device 152 and the first battery cell supporting device 132 can be integrated into one. When the target battery cell 200 is transferred on the battery cell supporting device to the bottom of the first image acquisition device 136, the transfer is stopped. After the first image acquisition device 136 completes the single spectrum image acquisition of the target battery cell, the target battery cell is continued to be transferred to the bottom of the second image acquisition device 156 to collect three-dimensional point cloud data.
[0167] The second battery cell supporting device 152 carries a target battery cell 200, which is currently located at a second inspection station. The second inspection station is configured to collect at least three-dimensional point cloud data about the target battery cell 200. As shown in FIG14 , during the three-dimensional point cloud data collection, the target battery cell 200 moves relative to the second light source 154 and the second image acquisition device 156 in the direction indicated by the arrow in the figure.
[0168] In some embodiments, the second light source 154 includes a laser generator; the second battery cell carrier 152 moves the target battery cell 200 in a predetermined manner and passes under the second light source 154 in a predetermined movement mode and / or based on instructions from a computing device; the second image acquisition device 156 is arranged in the upper area of the second battery cell carrier 152 and is at a predetermined angle to the laser generator so that the second image acquisition device 156 receives the laser reflected from the top surface of the target battery cell 200.
[0169] Figure 15 shows a schematic step diagram of an example electronic device 1500 that can be used to implement an embodiment of the present disclosure. For example, the computing device 110 as shown in Figure 1 can be implemented by the electronic device 1500. As shown in the figure, the electronic device 1500 includes a central processing unit (CPU) 1501, which can perform various appropriate actions and processes according to the computer program instructions stored in a read-only memory (ROM) 1502 or the computer program instructions loaded into a random access memory (RAM) 1503 from a storage unit 1508. In the random access memory 1503, various programs and data required for the operation of the electronic device 1500 can also be stored. The central processing unit 1501, the read-only memory 1502 and the random access memory 1503 are connected to each other by a bus 1504. An input / output (I / O) interface 1505 is also connected to the bus 1504.
[0170] Multiple components in electronic device 1500 are connected to input / output interface 1505, including: an input unit 1506, such as a keyboard, mouse, microphone, etc.; an output unit 1507, such as various types of displays, speakers, etc.; a storage unit 1508, such as a magnetic disk, optical disk, etc.; and a communication unit 1509, such as a network card, modem, wireless communication transceiver, etc. The communication unit 1509 allows device 1500 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.
[0171] The various processes and procedures described above, such as methods 300 to 900, may be performed by the central processing unit 1501. For example, in some embodiments, methods 300 to 900 may be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit 1508. In some embodiments, part or all of the computer program may be loaded and / or installed onto the device 1500 via the read-only memory 1502 and / or the communication unit 1509. When the computer program is loaded into the random access memory 1503 and executed by the central processing unit 1501, one or more actions of the methods 300 to 900 described above may be performed.
[0172] The present disclosure relates to methods, apparatuses, systems, electronic devices, computer-readable storage media, and / or computer program products. The computer program products may include computer-readable program instructions for executing various aspects of the present disclosure.
[0173] A computer-readable storage medium can be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination thereof. As used herein, a computer-readable storage medium is not to be construed as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through an electrical wire.
[0174] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge computing devices. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.
[0175] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages such as Smalltalk, C++, and conventional procedural programming languages such as "C" language or similar programming languages. Computer-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., utilizing an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be personalized by utilizing the state information of the computer-readable program instructions. The electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.
[0176] Various aspects of the present disclosure are described herein with reference to flowcharts and / or step diagrams of methods, devices (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each step in the flowcharts and / or step diagrams, and any combination of the steps in the flowcharts and / or step diagrams, can be implemented by computer-readable program instructions.
[0177] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine such that when these instructions are executed by the processing unit of the computer or other programmable data processing device, a device is generated that implements the functions / actions specified in one or more steps in the flowchart and / or step diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium, where these instructions cause the computer, programmable data processing device, and / or other device to operate in a specific manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more steps in the flowchart and / or step diagram.
[0178] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more steps in the flowchart and / or step diagram.
[0179] The flowcharts and step diagrams in the accompanying drawings show the possible architectures, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each step in the flowchart or step diagram can represent a module, program segment or part of an instruction, and a module, program segment or part of an instruction contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions marked in the steps can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each step in the step diagram and / or flowchart, and the combination of the steps in the step diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.
[0180] While various embodiments of the present disclosure have been described above, the foregoing description is intended to be illustrative, non-exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or technological improvements in the marketplace, or to enable others skilled in the art to understand the embodiments disclosed herein.
Claims
1. A defect detection method for a battery core explosion-proof valve, characterized in that: include: Acquire a plurality of single spectrum images of different wavelengths about a target battery cell, each single spectrum image at least indicating an explosion-proof valve region of the target battery cell; fusing a plurality of single-spectrum images of different wavelengths to generate a multi-spectral fused image of the target battery cell; and Based on the generated multi-spectral fusion image, a defect detection result of the explosion-proof valve of the target battery cell is determined.
2. The method according to claim 1, characterized in that Acquiring single spectrum images of multiple different wavelengths of the target battery cell includes: During a single exposure time, the light source of the first inspection station is controlled to switch between lights of multiple different wavelengths so that the first image acquisition device can acquire single-spectrum images of multiple different wavelengths about the target battery cell, wherein the station parameters of the first inspection station are related to the parameters of the first image acquisition device, the light source parameters and / or the position of the target battery cell.
3. The method according to claim 2, characterized in that The controlling of the light source of the first detection station to switch to a plurality of lights of different wavelengths comprises: The light source of the first detection station switches the multiple lights of different wavelengths in a time-sharing stroboscopic manner; The wavelength of any of the light is greater than or equal to the wavelength of ultraviolet light and less than or equal to the wavelength of infrared light.
4. The method according to claim 2 or 3, characterized in that: The controlling of the light source of the first detection station to switch to a plurality of lights of different wavelengths comprises: Switch the ultraviolet light, blue light, green light, orange light, red light, far infrared light, infrared light and white light sources in sequence.
5. The method according to claim 3, characterized in that: Also includes: The wavelength of each of the plurality of different wavelengths of light is determined based on the type of defect to be detected.
6. The method according to claim 4, characterized in that Fusion of multiple single-spectrum images at different wavelengths includes: Fusion of a single spectrum image of ultraviolet wavelength, a single spectrum image of blue light wavelength, a single spectrum image of green light wavelength, a single spectrum image of orange light wavelength, a single spectrum image of red light wavelength and a single spectrum image of white light wavelength of the target battery cell; and / or The single spectrum image of the target battery cell at the infrared wavelength and the single spectrum image at the far-infrared wavelength are fused.
7. The method according to any one of claims 1 to 6, characterized in that Determining a defect detection result of the explosion-proof valve of the target battery cell includes: Based on the generated multi-spectral fusion image, a first defect detection result of the explosion-proof valve of the target battery cell is determined, and the first defect at least indicates a planar defect of the explosion-proof valve of the target battery cell.
8. The method according to claim 7, characterized in that Determining a first defect detection result of the explosion-proof valve of the target battery cell includes: Based on a predetermined detection model, feature extraction is performed on the generated multi-spectral fusion image to determine defective pixel information of the area where the explosion-proof valve of the target battery cell is located, wherein the defective pixel information indicates defective pixel category information and Corresponding defective pixel location information; determining a size of a defective region associated with the defective pixel category information based on the defective pixel category information and the corresponding defective pixel location information; and In response to determining that the size of the defect area is greater than a predetermined defect area size threshold, determining that a first defect detection result regarding the explosion-proof valve of the target battery cell indicates that the explosion-proof valve of the target battery cell has a planar defect.
9. The method according to claim 8, characterized in that The planar defects include at least one of the following: electrolyte sputtering, corrosion, dirt, and foreign matter.
10. The method according to claim 1, characterized in that Fusion of multiple single-spectrum images at different wavelengths includes: In response to determining that a characteristic parameter of any single-spectrum image among the plurality of single-spectrum images of different wavelengths meets a predetermined characteristic parameter threshold condition, extracting the any single-spectrum image; and Image fusion is performed based on the features of the extracted single-spectrum image to generate a multi-spectral fusion image of the target battery cell.
11. The method according to claim 7, characterized in that Also includes: Acquire three-dimensional point cloud data about the target battery cell; as well as Based on the acquired three-dimensional point cloud data, the height information of the top surface of the target battery cell is calculated so as to determine the second defect detection result of the explosion-proof valve of the target battery cell based on the height information, and the second defect at least indicates a morphological defect of the explosion-proof valve of the target battery cell.
12. The method according to claim 11, characterized in that Calculating the height information of the top surface of the target battery cell includes: In response to a first defect detection result of the explosion-proof valve of the target battery cell indicating the presence of a predetermined defect type, height information of a top surface of the target battery cell is calculated based on the acquired three-dimensional point cloud data.
13. The method according to claim 11, characterized in that Determining a second defect detection result of the explosion-proof valve of the target battery cell based on the height information includes: Determine a mapping relationship between the acquired three-dimensional point cloud data and the generated multi-spectral fusion image, so as to determine a region of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data, wherein the region of interest includes at least a portion of the region where the explosion-proof valve is located; Performing plane fitting on the determined region of interest so as to calculate height information of the region of interest; and Based on the calculated height information of the region of interest, it is determined whether the second defect detection result of the explosion-proof valve of the target battery cell indicates that the explosion-proof valve of the target battery cell has a morphological defect.
14. The method according to claim 13, characterized in that Determining the mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image includes: Preprocessing the acquired three-dimensional point cloud data by bilateral filtering to remove abnormal data in the acquired three-dimensional point cloud data; Extracting a plurality of first key points from the three-dimensional point cloud data after removing abnormal data, so as to determine a spatial conversion relationship between the acquired three-dimensional point cloud data and the generated multi-spectral fusion image; and Based on the determined spatial conversion relationship, a mapping relationship between the acquired three-dimensional point cloud data and the generated multispectral fusion image is determined.
15. The method according to claim 14, characterized in that Determining the region of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data includes: Based on the first key point, the region of interest is determined, and the region of interest at least includes a region where a first defect of the explosion-proof valve of the target battery cell is located.
16. The method according to claim 14, characterized in that Determining the region of interest on the top surface of the target battery cell in the acquired three-dimensional point cloud data further includes: Extracting a plurality of second key points in the area where the explosion-proof valve of the target battery cell is located, and / or determining a plurality of second key points in the first key points; Determining a region where a first defect of the explosion-proof valve of the target battery cell is located based on the second key point; and In the acquired three-dimensional point cloud data, the area where the first defect is located is determined as the area of interest.
17. The method according to claim 13, characterized in that The morphological defects include at least one of the following: breakage and deformation.
18. A computing device, characterized in that include: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 17.
19. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 17.
20. A computer program product, characterized in that A computer program is included which, when executed by a processor, implements the method according to any one of claims 1 to 17.
21. A defect detection system for a battery core explosion-proof valve, characterized in that: include: A battery cell carrying device, configured to carry a target battery cell and having at least one inspection station; At least one light source is configured to respectively emit light of a plurality of different wavelengths for irradiating a target battery cell and emit laser light for irradiating the target battery cell; At least one image acquisition device is configured to acquire single-spectrum images of multiple different wavelengths of the target battery cell and to acquire three-dimensional point cloud data of the target battery cell at the at least one inspection station; The computing device according to claim 19 is configured to determine a defect detection result of the explosion-proof valve of the target battery cell based on a plurality of single-spectrum images of different wavelengths acquired by at least one image acquisition device.
22. The system according to claim 21, characterized in that The at least one light source includes a plurality of LED sub-light sources that are ring-shaped and evenly distributed, and the at least one light source is configured to switch between at least two different wavelengths of light based on instructions from a computing device.
Citation Information
Patent Citations
Rod and wire surface defect online detection device and method
CN108490000A
Product surface defect detection system based on multi-spectral imaging
CN110441312A
Surface defect automatic optical detection system and method based on multispectral spectroscopic imaging
CN111830046A
Image fusion system based on multispectral camera
CN114119443A
Unified neural network for defect detection and classification
US20190073568A1
Cited By
Battery defect detection method and system based on AI
CN119671970A
Cord fabric production defect identification and analysis system based on machine vision
CN120219368A
Battery monomer appearance detection device and detection method
CN120334235A
On-line quality detection method for nerve-calming and brain-nourishing liquid based on multispectral detection technology
CN120446023A
Stamping part detection method and system, electronic equipment and storage medium
CN120629156A
