Method and device for detecting invisible defects of crystalline silicon solar cell panel based on snapshot spectral imaging

Abstract

The invention provides a method and device for detecting invisible defects of a silicon crystal solar cell panel based on snapshot type spectral imaging, and the method comprises the steps: controlling a snapshot type spectral imaging device and an RGB camera assembly, and collecting a full-wave-band multispectral image and a visible light image for the silicon crystal solar cell panel in a high-irradiance environment; through a pre-trained invisible defect detection model, the full-band multispectral image and the visible light image are mapped to the same feature space, and sunlight interference of a high-irradiance environment on the full-band multispectral image is eliminated by using the visible light image. Performing interactive processing on the full-band multispectral image and the visible light image after the sunlight interference is eliminated in a feature space to obtain invisible defect features of the surface of the crystalline silicon solar cell panel, and determining an invisible defect detection result graph of the crystalline silicon solar cell panel based on the invisible defect features. According to the invention, the limitation of a detection environment is broken through, and the dependence on external equipment is reduced, so that the reliability and applicability of detection are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and in particular to a method and device for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging. Background Art

[0002] Silicon crystal solar panel components are key devices in solar power generation systems. Due to various reasons, various invisible defects often exist inside the solar panel components, including corner defects, hidden cracks, fragments, black spots, broken grids, surface contamination, local heating, etc. These defects are important factors affecting the photoelectric conversion efficiency and service life of solar panel components, and even pose safety problems. Therefore, it is necessary to detect these invisible defects during the use of solar panel components to eliminate potential hazards. Currently, the online detection technologies for invisible defects of solar panels include infrared thermal imaging, electroluminescence imaging, photoluminescence imaging, I-V curve characteristic detection and analysis, etc. Among them, electroluminescence imaging and photoluminescence imaging technologies have the advantages of high detection sensitivity, high detection efficiency, and the ability to detect a variety of defect types, and have been widely used in the field of defect detection of solar panels. The so-called electroluminescence is to apply a certain reverse voltage to the PN junction of a photovoltaic panel, thereby inducing it to emit light of a certain wavelength; while photoluminescence is to irradiate the semiconductor material of a photovoltaic panel with light of a certain wavelength, thereby exciting it to emit light of a certain wavelength. The spectral energy of electroluminescence or photoluminescence of silicon crystal materials is mainly concentrated in the short-wave infrared range of wavelengths from 1050nm to 1250nm; against the background of this electroluminescence or photoluminescence, various invisible defects of solar panels are revealed, thus creating conditions for imaging. However, compared with the daylight radiation intensity on the ground during the day, the radiation intensity of electroluminescence or photoluminescence of solar panels is very limited. Therefore, in order to avoid interference from ambient daylight, most of the existing electroluminescence imaging and photoluminescence imaging technologies are used in indoor or darkroom environments. However, most of the existing solar panel components operate in outdoor daylight environments. Therefore, it is necessary to develop electroluminescence and photoluminescence imaging technologies under high irradiance during the day.

[0003] The main drawback of the prior art lies in its strict requirements for the detection environment. Since solar panels usually operate in outdoor high-irradiance environments, and the prior art lacks the ability to work effectively under daylight conditions, it is difficult to achieve rapid outdoor detection, which significantly limits the on-line detection of solar panel components in practical applications. In addition, traditional photoluminescence imaging technology relies on high-power excitation light sources, which not only makes the equipment complex and costly, but may also affect the detection ability of deep or weak defects due to insufficient excitation; electroluminescence imaging technology relies on external power supply, with complex operation and is not suitable for large-area detection scenarios, resulting in low efficiency. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging, which breaks through the limitations of the detection environment and reduces the dependence on external devices, thereby greatly improving the reliability and applicability of detection.

[0005] In the first aspect, the present invention provides a method for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging. The method is applied to a processing terminal of an invisible defect detection device. The invisible defect detection device further includes a snapshot spectral imaging device and an RGB camera assembly communicatively connected to the processing terminal. The method includes:

[0006] Controlling the snapshot spectral imaging device and the RGB camera assembly to collect full-band multi-spectral images and visible light images for the silicon crystal solar panel in a high-irradiance environment;

[0007] Through a pre-trained invisible defect detection model, mapping the full-band multi-spectral images and visible light images to the same feature space, using the visible light images to eliminate the daylight interference of the high-irradiance environment on the full-band multi-spectral images, and performing interactive processing on the full-band multi-spectral images and visible light images after eliminating daylight interference in the feature space to obtain invisible defect features on the surface of the silicon crystal solar panel, and determining an invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features.

[0008] In one embodiment, mapping the full-band multi-spectral images and visible light images to the same feature space includes:

[0009] Respectively extracting the full-band feature information of the full-band multi-spectral images and the illumination feature information of the visible light images, mapping the full-band feature information and the illumination feature information to the same feature space, and aligning the full-band feature information and the illumination feature information in the feature space.

[0010] In one embodiment, using the visible light images to eliminate the daylight interference of the high-irradiance environment on the full-band multi-spectral images includes:

[0011] Perform a difference operation on the full-band feature information and the illumination feature information to pixel-by-pixel remove the illumination feature information from the full-band feature information, obtaining new full-band feature information, and realizing the elimination of the sunlight interference existing in the full-band multispectral image in a high irradiance environment.

[0012] In one implementation, perform interactive processing on the full-band multispectral image after eliminating sunlight interference and the visible light image in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, including:

[0013] Perform channel attention processing on the new full-band feature information and the illumination feature information;

[0014] Generate target feature information based on the full-band feature information after channel attention processing and the illumination feature information after channel attention processing;

[0015] Extract the invisible defect features on the surface of the silicon crystal solar panel from the target feature information.

[0016] In one implementation, generating target feature information based on the full-band feature information after channel attention processing and the illumination feature information after channel attention processing includes:

[0017] Perform splicing on the full-band feature information after channel attention processing and the illumination feature information after channel attention processing, and assign weight values to the full-band feature information after channel attention processing and the illumination feature information after channel attention processing to obtain target feature information;

[0018] Among them, the weight value assigned to the full-band feature information after channel attention processing is higher than the weight value assigned to the illumination feature information after channel attention processing.

[0019] In one implementation, determining the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features includes:

[0020] Map the invisible defect features to the full-band multispectral image to obtain a new full-band multispectral image;

[0021] Based on the spectral characteristics of the new full-band multispectral image, identify the invisible defects existing in the silicon crystal solar panel to generate an invisible defect detection result map.

[0022] In a second aspect, the present invention also provides an invisible defect detection device for a silicon crystal solar panel based on snapshot spectral imaging, including a processing terminal, and a snapshot spectral imaging device and an RGB camera assembly communicatively connected to the processing terminal; the processing terminal includes:

[0023] A control module for controlling a snapshot spectral imaging device and an RGB camera assembly to collect full-band multi-spectral images and visible light images of a silicon crystal solar panel in a high irradiance environment.

[0024] A defect detection module for mapping the full-band multi-spectral image and the visible light image to the same feature space through a pre-trained invisible defect detection model, eliminating the sunlight interference of the high irradiance environment on the full-band multi-spectral image and the visible light image, and performing interactive processing on the full-band multi-spectral image and the visible light image after eliminating the sunlight interference in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, and determining the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features.

[0025] In one implementation, the defect detection module is specifically used for:

[0026] Respectively extract the full-band feature information of the full-band multi-spectral image and the illumination feature information of the visible light image, map the full-band feature information and the illumination feature information to the same feature space, and align the full-band feature information and the illumination feature information in the feature space.

[0027] In a third aspect, the present invention further provides a processing terminal, including a processor and a memory, the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method of any one provided in the first aspect.

[0028] In a fourth aspect, the present invention further provides a computer-readable storage medium, the computer-readable storage medium stores computer executable instructions, and when the computer executable instructions are called and executed by the processor, the computer executable instructions cause the processor to implement the method of any one provided in the first aspect.

[0029] A method and device for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging provided by the present invention first control a snapshot spectral imaging device and an RGB camera assembly to collect full-band multi-spectral images and visible light images of the silicon crystal solar panel in a high irradiance environment; then, through a pre-trained invisible defect detection model, map the full-band multi-spectral images and visible light images to the same feature space, use the visible light images to eliminate the sunlight interference of the high irradiance environment on the full-band multi-spectral images, and perform interactive processing on the full-band multi-spectral images and visible light images after eliminating the sunlight interference in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, and determine the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features. The above method uses a snapshot spectral imaging device to collect full-band multi-spectral images of the silicon crystal solar panel in a high irradiance environment. After mapping the full-band multi-spectral images and visible light images to the same feature space, use the visible light images to eliminate the sunlight interference of the high irradiance environment on the full-band multi-spectral images. On this basis, perform interactive processing on the full-band multi-spectral images and visible light images in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, so as to generate invisible defect detection features. The present invention breaks through the limitation of the detection environment and reduces the dependence on external devices, thereby greatly improving the reliability and applicability of the detection.

[0030] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention are achieved and obtained by the structures specifically pointed out in the specification, claims, and drawings.

[0031] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specifically provides preferred embodiments and, in conjunction with the accompanying drawings, the detailed description is as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0033] Figure 1 It is a schematic flow chart of a method for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging provided by an embodiment of the present invention;

[0034] Figure 2 It is a schematic structural diagram of a snapshot spectral imaging device provided by an embodiment of the present invention;

[0035] Figure 3 This is a technical framework diagram of a method for detecting invisible defects in silicon crystal solar panels based on snapshot spectral imaging provided by an embodiment of the present invention;

[0036] Figure 4 This is a schematic structural diagram of an invisible defect detection model provided by an embodiment of the present invention;

[0037] Figure 5 This is a schematic software structure diagram of a processing terminal provided by an embodiment of the present invention;

[0038] Figure 6 This is a schematic hardware structure diagram of a processing terminal provided by an embodiment of the present invention. Detailed implementation manners

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] The prior art has strict requirements for the detection environment, and the detection effect highly depends on external devices. For example, CN118199517A "Photoluminescence imaging system and method for photovoltaic panels based on a linear array InGaAs camera", CN201340393Y "On-line detection device for defects of reflective solar cell modules", WO2011152445A1 "Electroluminescence detection device and electroluminescence detection method for solar panels", and WO2017172611A1 "Automatic identification and defect detection system and method for solar panels using infrared imaging" do not involve imaging detection in daylight environments.

[0041] Although some literatures attempt to study defect detection under high illumination conditions, they do not involve combining multi-spectral imaging technology to improve anti-interference ability. For example, the literatures "Research on defect detection technology for photovoltaic panels based on a high frame rate InGaAs camera" and "Research on defect detection technology for solar panels under high illumination based on an InGaAs camera" do not involve multi-spectral imaging problems.

[0042] In addition, the prior art also lacks precise classification and identification means for complex defects, further restricting the wide range and reliability of its practical applications.

[0043] Based on this, the embodiments of the present invention provide a method and device for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging, which break through the limitations of the detection environment and reduce the dependence on external devices, thereby greatly improving the reliability and applicability of detection.

[0044] For the convenience of understanding this embodiment, first, a method for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging disclosed in the embodiments of the present invention will be introduced in detail. This method is applied to the processing terminal of the invisible defect detection device. The invisible defect detection device further includes a snapshot spectral imaging device and an RGB camera assembly that are communicatively connected to the processing terminal. Refer to Figure 1 the schematic flowchart of a method for detecting invisible defects of silicon crystal solar panels based on snapshot spectral imaging as shown. This method mainly includes the following steps S102 to S104:

[0045] Step S102, control the snapshot spectral imaging device and the RGB camera assembly to collect full-band multi-spectral images and visible light images of the silicon crystal solar panel in a high-irradiance environment.

[0046] In one example, the processing terminal can send control instructions to the snapshot spectral imaging device and the RGB camera assembly respectively to trigger the snapshot spectral imaging device to collect full-band multi-spectral images of the silicon crystal solar panel in a high-irradiance environment and trigger the RGB camera assembly to collect visible light images of the silicon crystal solar panel in a high-irradiance environment.

[0047] Step S104, through a pre-trained invisible defect detection model, map the full-band multi-spectral image and the visible light image to the same feature space, use the visible light image to eliminate the sunlight interference of the full-band multi-spectral image in the high-irradiance environment, and perform interactive processing on the full-band multi-spectral image and the visible light image after eliminating the sunlight interference in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, and determine the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features.

[0048] Among them, the invisible defect detection result map is used to describe the position of invisible defects on the silicon crystal solar panel, and can also mark the defect types to which each invisible defect belongs, such as tiny cracks, fine surface contamination, local oxidation, etc. In one example, image mapping can be used to map the full-band multi-spectral image and the visible light image to the same feature space; differential processing is used to eliminate the daylight interference of the full-band multi-spectral image in a high irradiance environment by using the visible light image; after splicing and assigning weight values to the full-band multi-spectral image and the visible light image after eliminating daylight interference, the invisible defect features on the surface of the silicon crystal solar panel are extracted; finally, based on the invisible defect features and the spectral characteristics of the full-band multi-spectral image, the invisible defects on the surface of the silicon crystal solar panel are identified, and the invisible defect detection result map is output.

[0049] The method for detecting invisible defects on a silicon crystal solar panel based on snapshot spectral imaging provided by the embodiments of the present invention has at least the following advantages: (a) Environmental interference elimination: Through differential processing and interactive operations in the feature space, the algorithm can significantly reduce the environmental daylight interference in strong light environments and improve the detection ability under complex lighting conditions. (b) Accurate identification of invisible defects: Through high-resolution image reconstruction and feature interaction, the algorithm can identify tiny and imperceptible invisible defects, including surface cracks, local contamination, etc. (c) Efficient processing: The model is optimized to be able to process a large amount of image data in a short time, suitable for large-scale panel detection, and can maintain high-efficiency and accurate detection performance.

[0050] For ease of understanding, first, the device for detecting invisible defects on a silicon crystal solar panel based on snapshot spectral imaging is explained. The device includes a processing terminal, and a snapshot spectral imaging device and an RGB camera assembly communicatively connected to the processing terminal. Among them, the snapshot spectral imaging device uses an advanced pixel-level spectroscopic imaging device. Refer to Figure 2 the structural schematic diagram of a snapshot spectral imaging device shown. This device can instantaneously capture the full-band multi-spectral image of the photoluminescence or electroluminescence of the solar panel, and then perform fusion processing on the collected full-band multi-spectral image, further improving the imaging resolution of the full-band multi-spectral image while overcoming the interference of environmental daylight, so as to realize the efficient detection and identification of invisible defects on the solar panel under high irradiance.

[0051] Compared with traditional defect detection techniques that usually rely on single - band imaging methods such as infrared thermal imaging, electroluminescence imaging (EL), and photoluminescence imaging (PL), the detection capabilities of these methods are affected by environmental light changes and it is difficult to provide sufficient image resolution. In contrast, in the embodiments of the present invention, by innovatively applying snapshot spectral imaging technology, it is possible to simultaneously obtain high - resolution multi - spectral images within the full - band range of a solar panel during a single shooting process with the assistance of an imaging algorithm. This technology significantly improves the detection accuracy. Especially in a high - irradiance environment, it can overcome environmental sunlight interference, break through the limitations of the detection environment, and greatly improve the reliability and applicability of defect detection.

[0052] On this basis, the embodiments of the present invention provide a specific implementation of a method for detecting invisible defects in silicon crystal solar panels based on snapshot spectral imaging. Refer to Figure 3 the technical framework diagram of a method for detecting invisible defects in silicon crystal solar panels based on snapshot spectral imaging shown in

[0053] This method combines the processing technology of full - band multi - spectral images, can effectively extract the invisible defect features on the surface of the solar panel in a high - irradiance environment, and optimizes the detection accuracy through a deep learning model. Compared with traditional methods, this method can more accurately identify invisible defects that are difficult to observe, improves the sensitivity and accuracy of defect detection, and significantly enhances the reliability and adaptability of this detection device in practical applications.

[0054] In a specific implementation, refer to Figure 4 the structural schematic diagram of an invisible defect detection model shown in

[0055] (1) Map the full-band multispectral image and the visible light image to the same feature space. Specifically: Extract the full-band feature information of the full-band multispectral image and the illumination feature information of the visible light image respectively, map the full-band feature information and the illumination feature information to the same feature space, and align the full-band feature information and the illumination feature information in the feature space.

[0056] Map the visible light image and the full-band multispectral image to the same feature space. Since different types of images (such as RGB visible light images and multispectral images) usually have different resolutions, spectral characteristics, and noise characteristics, separate processing may lead to information mismatch. Therefore, through image mapping, these images are converted into a unified feature space so that they can be processed subsequently in a consistent manner.

[0057] (2) Use the visible light image to eliminate the sunlight interference of the full-band multispectral image in a high-irradiance environment. Specifically: Perform a differential operation on the full-band feature information and the illumination feature information to remove the illumination feature information from the full-band feature information pixel by pixel, obtaining new full-band feature information, and realizing the elimination of the sunlight interference of the full-band multispectral image in a high-irradiance environment.

[0058] Reduce the interference caused by environmental sunlight through differential operation. The differential operation is to calculate the pixel difference between the visible light image and the full-band multispectral image, remove the background illumination change, and retain the details of the battery panel surface. This process effectively reduces the image noise and interference caused by environmental illumination change in a strong light environment, especially under sunlight irradiation, so that the information of the battery panel surface and potential defects is clearer.

[0059] (3) Perform interactive processing on the full-band multispectral image and the visible light image after eliminating sunlight interference in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel. Specifically: Perform channel attention processing on the new full-band feature information and the illumination feature information; Generate target feature information based on the full-band feature information after channel attention processing and the illumination feature information after channel attention processing; Extract the invisible defect features on the surface of the silicon crystal solar panel from the target feature information.

[0060] In one example, the process of generating target feature information is as follows: Concatenate the full-band feature information after channel attention processing and the illumination feature information after channel attention processing, and assign weight values to the full-band feature information after channel attention processing and the illumination feature information after channel attention processing to obtain target feature information.

[0061] After image mapping and differential processing, it enters the feature space processing stage. In this stage, the model processes the information in the mapped feature space interactively. The purpose of feature interaction is to fully integrate the information from different spectral bands to enhance the manifestation of defects.

[0062] Spectral images of different bands provide different characterization capabilities. For example, the infrared band may be more sensitive to surface cracks, while the visible light band may have a better response to contamination or oxidation. Through feature interaction, the model can effectively integrate these useful information from different bands in the feature space, further reducing noise and interference and highlighting subtle defect features. For example, for smaller invisible cracks, the interactive processing can strengthen the feature information related to the defects in different bands, making these defects more prominent in the image.

[0063] The key to this process is to preferentially process the most discriminative features in the multispectral image through feature selection and weighting strategies, that is, the weight value given to the full-band feature information after channel attention processing is higher than the weight value given to the illumination feature information after channel attention processing, so as to improve the sensitivity and accuracy of defect detection.

[0064] (4) Determine the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features. Specifically, map the invisible defect features to the full-band multispectral image to obtain a new full-band multispectral image, and identify the invisible defects existing in the silicon crystal solar panel based on the spectral characteristics of the new full-band multispectral image to generate an invisible defect detection result map.

[0065] After completing the interactive processing in the feature space, the algorithm maps the processed information back to the high-resolution multispectral image for defect detection. The process of defect detection is as follows: use the convolutional decoding network to gradually restore the resolution of the feature map, and adopt a classification prediction head composed of a two-dimensional convolutional layer and a Sigmoid activation function to classify the fused features to generate the final defect detection result map. This step mainly identifies potential invisible defects on the surface and inside of the solar panel by analyzing the spectral characteristics of each band in the multispectral image.

[0066] The high-resolution image provides more detailed spatial information, enabling the detection device to be accurate to tiny cracks, subtle surface contamination, local oxidation, etc. These defects are often difficult to detect by traditional methods. By deeply analyzing the spectral characteristics, the algorithm can capture the reflectance changes, luminescence characteristics and other relevant information in different bands, and these features are often directly related to the defects of the solar panel. Finally, the algorithm gives a high-precision defect detection result through comprehensive analysis of all bands.

[0067] Further, an embodiment of the present invention provides a specific application example of a method for detecting invisible defects in silicon crystal solar panels based on snapshot spectral imaging, including:

[0068] The first step: Under natural sunlight irradiation, the interior and surface of the solar panel assembly reflect visible light and infrared light. The visible light is imaged by a high-resolution RGB camera to generate a high-resolution RGB image. At the same time, a snapshot spectral imager is used to capture the full-band spectral image and record the full-band spectral information of the solar panel. All image data are initially grayscaled, normalized, and noise-filtered by an image processing unit to ensure image quality.

[0069] The second step: The visible light RGB image and the full-band multispectral image are mapped to the same feature space. First, differential operations are performed to eliminate possible sunlight interference in a high-irradiance environment and remove image noise caused by strong light illumination. The differential image provides cleaner information about the surface of the panel, removing the influence of background light.

[0070] The third step: Feature interaction processing is carried out within the feature space. The algorithm strengthens defect information according to the spectral characteristics of different bands. Through feature interaction, information from the visible light and multispectral bands is effectively fused, making subtle defects (such as microcracks, local contamination, etc.) more prominent and enhancing the expressiveness of defects.

[0071] The fourth step: After completing the processing in the feature space, the processed feature information is mapped back to the high-resolution multispectral image for defect detection. The process of defect detection is as follows: The resolution of the feature map is gradually restored using a convolutional decoding network, and a classification prediction head composed of a two-dimensional convolutional layer and a Sigmoid activation function is used to classify the fused features to generate the final defect detection result map. By analyzing the spectral characteristics in the image (such as reflectance changes, photoluminescence, or electroluminescence characteristics), the algorithm can accurately identify invisible defects on the solar panel.

[0072] In summary, the embodiment of the present invention can achieve the following technical effects:

[0073] (a) Improve the detection accuracy under high-irradiance conditions: The embodiment of the present invention adopts a new snapshot spectral imaging technology, which can effectively capture the multispectral image of the solar panel in a high-irradiance environment, and through image fusion and feature interaction processing, significantly improve the detection ability of invisible defects. Compared with the prior art that relies on single-band imaging or is limited by traditional infrared imaging, the full-band imaging method of the embodiment of the present invention not only overcomes the interference of environmental sunlight but also improves the detection accuracy, and can accurately identify invisible defects such as microcracks and surface contamination.

[0074] (b) Effective suppression of environmental interference: Existing technologies are often affected by changes in environmental light, especially under strong sunlight, resulting in image noise and artifacts, which in turn affect the accuracy of defect detection. In contrast, the embodiments of the present invention can effectively eliminate sunlight interference in a high-irradiance environment through differential image processing and interactive processing in the feature space, retain more pure defect information, and make the detection results more reliable.

[0075] (c) Multi-spectral fusion to improve image resolution: The embodiments of the present invention combine visible light images and full-band multi-spectral images, and use image fusion technology to improve the resolution. Especially in full-band multi-spectral imaging, the details of the multi-spectral image are reconstructed under the guidance of a high-resolution RGB visible light image, making the defect details on the surface of the solar panel more prominent and further improving the accuracy of defect detection. In contrast, existing technologies usually lack such an integrated high-resolution multi-spectral image fusion technology.

[0076] (d) Efficient feature space interactive processing: The embodiments of the present invention effectively fuse information from different spectral bands through innovative feature space interactive processing, enhance the expression of defects, and make subtle invisible defects more obvious. This interactive processing method is more sensitive than traditional single-band detection methods and can detect defects that are difficult to detect by traditional methods, such as tiny cracks and oxide layers.

[0077] (e) Adapt to complex environmental conditions: Compared with traditional defect detection methods, the technical solutions of the embodiments of the present invention break through the limitations of traditional detection environments and can work stably under complex and strong light conditions without relying on additional shading devices or complex environmental adjustments. Therefore, this technology is more adaptable in practical applications, especially when widely used in the detection of large-scale solar panels, it can provide efficient and stable detection results.

[0078] Based on the foregoing embodiments, the embodiments of the present invention provide a silicon crystal solar panel invisible defect detection device based on snapshot spectral imaging. The device includes a processing terminal, and a snapshot spectral imaging device and an RGB camera assembly communicatively connected to the processing terminal. Among them, referring to Figure 5 the schematic software structure diagram of a processing terminal shown, the processing terminal includes the following parts:

[0079] A control module 502, configured to control the snapshot spectral imaging device and the RGB camera assembly to collect full-band multi-spectral images and visible light images of the silicon crystal solar panel under a high-irradiance environment;

[0080] The defect detection module 506 is used to map the full-band multispectral image and the visible light image to the same feature space through a pre-trained invisible defect detection model, eliminate the sunlight interference of the high irradiance environment on the full-band multispectral image and the visible light image, and perform interactive processing on the full-band multispectral image and the visible light image after eliminating the sunlight interference in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, and determine the invisible defect detection result map of the silicon crystal solar panel based on the invisible defect features.

[0081] The invisible defect detection device for silicon crystal solar panels based on snapshot spectral imaging provided by the embodiments of the present invention uses a snapshot spectral imaging device to collect the full-band multispectral image of the silicon crystal solar panel in a high irradiance environment. After mapping the full-band multispectral image and the visible light image to the same feature space, the visible light image is used to eliminate the sunlight interference of the high irradiance environment on the full-band multispectral image. On this basis, the full-band multispectral image and the visible light image are interactively processed in the feature space to obtain the invisible defect features on the surface of the silicon crystal solar panel, so as to generate invisible defect detection features. The present invention breaks through the limitation of the detection environment and reduces the dependence on external devices, thereby greatly improving the reliability and applicability of the detection.

[0082] In one embodiment, the defect detection module 506 is specifically used for:

[0083] Extract the full-band feature information of the full-band multispectral image and the illumination feature information of the visible light image respectively, map the full-band feature information and the illumination feature information to the same feature space, and align the full-band feature information and the illumination feature information in the feature space.

[0084] In one embodiment, the defect detection module 506 is specifically used for:

[0085] Perform a difference operation on the full-band feature information and the illumination feature information to eliminate the illumination feature information from the full-band feature information pixel by pixel, and obtain new full-band feature information, so as to eliminate the sunlight interference of the high irradiance environment on the full-band multispectral image.

[0086] In one embodiment, the defect detection module 506 is specifically used for:

[0087] Perform channel attention processing on the new full-band feature information and the illumination feature information;

[0088] Generate target feature information based on the full-band feature information after channel attention processing and the illumination feature information after channel attention processing;

[0089] Extract the invisible defect features on the surface of the silicon crystal solar panel from the target feature information.

[0090] In one embodiment, the defect detection module 506 is specifically configured to:

[0091] Concatenate the full-band feature information after channel attention processing and the illumination feature information after channel attention processing, and assign weight values to the full-band feature information after channel attention processing and the illumination feature information after channel attention processing to obtain target feature information;

[0092] Among them, the weight value assigned to the full-band feature information after channel attention processing is higher than the weight value assigned to the illumination feature information after channel attention processing.

[0093] In one embodiment, the defect detection module 506 is specifically configured to:

[0094] Map the invisible defect features to the full-band multispectral image to obtain a new full-band multispectral image;

[0095] Based on the spectral characteristics of the new full-band multispectral image, identify the invisible defects existing in the silicon crystal solar panel to generate an invisible defect detection result map.

[0096] The device provided by the embodiment of the present invention has the same implementation principle and the same technical effects as those of the foregoing method embodiment. For the sake of brief description, for the parts not mentioned in the device embodiment, reference may be made to the corresponding content in the foregoing method embodiment.

[0097] The embodiment of the present invention provides a processing terminal. Specifically, the processing terminal includes a processor and a storage device; a computer program is stored on the storage device, and the computer program executes the method according to any one of the foregoing embodiments when being run by the processor.

[0098] Figure 6 FIG. is a schematic hardware structure diagram of a processing terminal provided by an embodiment of the present invention. The processing terminal 100 includes: a processor 60, a memory 61, a bus 62, and a communication interface 63. The processor 60, the communication interface 63, and the memory 61 are connected through the bus 62; the processor 60 is used to execute an executable module stored in the memory 61, such as a computer program.

[0099] Among them, the memory 61 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory, such as at least one disk memory. Through at least one communication interface 63 (which may be wired or wireless), a communication connection between the system network element and at least one other network element is realized, and the Internet, a wide area network, a local area network, a metropolitan area network, etc. can be used.

[0100] The bus 62 can be an ISA bus, a PCI bus, an EISA bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 6 only a bidirectional arrow is used in Figure 6 , but it does not mean that there is only one bus or one type of bus.

[0101] Among them, the memory 61 is used to store a program. After receiving an execution instruction, the processor 60 executes the program. The method executed by the device defined by the flow process disclosed in any embodiment of the foregoing embodiments of the present invention can be applied to or implemented by the processor 60.

[0102] The processor 60 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit in the hardware of the processor 60 or the instructions in the form of software. The above-mentioned processor 60 may be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it may also be a digital signal processor (DSP for short), an application specific integrated circuit (ASIC for short), a field-programmable gate array (FPGA for short), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present invention can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 61, and the processor 60 reads the information in the memory 61 and combines its hardware to complete the steps of the above method.

[0103] The computer program product of the readable storage medium provided by the embodiments of the present invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the method described in the foregoing method embodiments. For the specific implementation, reference can be made to the foregoing method embodiments, and details are not described herein again.

[0104] When the above-mentioned functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0105] Finally, it should be noted that the above-mentioned embodiments are only specific embodiments of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions described in the foregoing embodiments or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes, or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for detecting invisible defects of silicon solar panels based on snapshot spectral imaging, characterized in that: The method is applied to a processing terminal of an invisible defect detection device, wherein the invisible defect detection device further comprises a snapshot spectral imaging device and an RGB camera assembly that are communicatively connected to the processing terminal. The method comprises: Controlling the snapshot spectral imaging device and the RGB camera assembly to collect full-band multispectral images and visible light images of silicon crystalline solar panels in a high irradiance environment; Through the pre-trained invisible defect detection model, the full-band multispectral image and the visible light image are mapped to the same feature space, and the visible light image is used to eliminate the sunlight interference of the high irradiance environment on the full-band multispectral image. The full-band multispectral image and the visible light image after eliminating the sunlight interference are interactively processed in the feature space to obtain the invisible defect features of the surface of the silicon crystalline solar panel, and the invisible defect detection result map of the silicon crystalline solar panel is determined based on the invisible defect features.

2. The method for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 1, characterized in that: Mapping the full-band multispectral image and the visible light image to the same feature space includes: The full-band feature information of the full-band multispectral image and the illumination feature information of the visible light image are extracted respectively, the full-band feature information and the illumination feature information are mapped to the same feature space, and the full-band feature information and the illumination feature information in the feature space are aligned.

3. The method for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 2 is characterized in that: Eliminating sunlight interference of the high irradiance environment on the full-band multispectral image by using the visible light image includes: A differential operation is performed on the full-band characteristic information and the illumination characteristic information to remove the illumination characteristic information from the full-band characteristic information pixel by pixel to obtain new full-band characteristic information, thereby eliminating the sunlight interference of the high irradiance environment on the full-band multispectral image.

4. The method for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 3 is characterized in that: Interactively processing the full-band multispectral image and the visible light image after eliminating the sunlight interference in the feature space to obtain invisible defect features on the surface of the silicon crystalline solar cell panel includes: Performing channel attention processing on the new full-band feature information and the illumination feature information; Generate target feature information based on the full-band feature information after the channel attention processing and the illumination feature information after the channel attention processing; The invisible defect features of the surface of the silicon crystalline solar cell panel are extracted from the target feature information.

5. The method for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 4, characterized in that: Generating target feature information based on the full-band feature information after the channel attention processing and the illumination feature information after the channel attention processing, including: splicing the full-band feature information after the channel attention processing and the illumination feature information after the channel attention processing, and assigning weight values ​​to the full-band feature information after the channel attention processing and the illumination feature information after the channel attention processing to obtain target feature information; Among them, the weight value assigned to the full-band feature information after the channel attention processing is higher than the weight value assigned to the illumination feature information after the channel attention processing.

6. The method for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 1, characterized in that: Determining the invisible defect detection result diagram of the silicon crystalline solar cell panel based on the invisible defect characteristics includes: Mapping the invisible defect features to the full-band multispectral image to obtain a new full-band multispectral image; Based on the spectral characteristics of the new full-band multi-spectral image, invisible defects existing in the silicon crystalline solar cell panel are identified to generate an invisible defect detection result map.

7. A device for detecting invisible defects of silicon solar panels based on snapshot spectral imaging, characterized in that: It includes a processing terminal, and a snapshot spectral imaging device and an RGB camera component that are communicatively connected to the processing terminal; the processing terminal includes: A control module, used to control the snapshot spectral imaging device and the RGB camera assembly to collect full-band multispectral images and visible light images of silicon crystalline solar panels in a high irradiance environment; The defect detection module is used to map the full-band multispectral image and the visible light image to the same feature space through a pre-trained invisible defect detection model, eliminate the sunlight interference of the high irradiance environment on the full-band multispectral image and the visible light image, and interactively process the full-band multispectral image and the visible light image after eliminating the sunlight interference in the feature space to obtain invisible defect features of the surface of the silicon crystalline solar cell panel, and determine the invisible defect detection result map of the silicon crystalline solar cell panel based on the invisible defect features.

8. The device for detecting invisible defects of silicon crystalline solar panels based on snapshot spectral imaging according to claim 7, characterized in that: The defect detection module is specifically used for: The full-band feature information of the full-band multispectral image and the illumination feature information of the visible light image are extracted respectively, the full-band feature information and the illumination feature information are mapped to the same feature space, and the full-band feature information and the illumination feature information in the feature space are aligned.

9. A processing terminal, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 6.

Images