Solar cell detection method and system

Abstract

The invention discloses a solar cell detection method and system, and the method comprises the steps: collecting multiple frames of solar cell images through a strip light source and an area-array camera, splicing the multiple frames of solar cell images to obtain a to-be-detected solar cell image, carrying out the gamma correction of the to-be-detected solar cell image, and obtaining a to-be-detected solar cell image. According to the invention, the strip light source and the area-array camera are used for acquisition, so that the cost of the camera and the light source is greatly reduced, the installation space of equipment is saved, and the resolution ratio of the area-array camera is low, so that the image detection time can be shortened, and the detection efficiency is improved. According to the method, the detection efficiency is improved, the image is optimized through image splicing and gamma correction, the uniformity of the overall brightness of the image is improved, the high detection rate and the low misjudgment rate can be guaranteed, the accuracy of the detection result is guaranteed, and therefore the economical efficiency, the high efficiency and the flexibility of detection are improved under the condition that the detection precision is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of visual inspection technology, and in particular to a solar cell inspection method and system. Background Art

[0002] In the photovoltaic industry, large-area defect detection is a key step to ensure the quality and performance of solar cells. Traditional solar cell inspection equipment mainly relies on high-resolution line scan cameras and complex image processing algorithms to achieve this goal. However, with the changes in market demand and technological advances, existing inspection methods have gradually shown some limitations, especially in terms of cost, efficiency and flexibility, as follows: (1) Conventional inspection equipment usually uses a line scan camera with a resolution of 5 megapixels (5MP) or higher for image acquisition. The working distance of this camera is generally 25 to 30 cm, and it can capture a complete image of the entire solar cell at one time. The line scan camera generates high-quality two-dimensional images by scanning the surface of the solar cell continuously at high speed, which are then transmitted to the computer system for analysis. Due to its high resolution characteristics, line scan cameras are very suitable for detecting tiny surface defects, cracks, stains, etc., as well as evaluating the uniformity and consistency of the cell. However, for large-area and wide-range defect detection, high-resolution line scan cameras have obvious limitations. First, the cost of high-resolution cameras is high, which increases the overall investment in equipment. Second, due to the need to process a large number of high-resolution images, computing resources are consumed, resulting in relatively slow inspection speed. Specifically, the image acquisition time and algorithm detection time are usually within 200 milliseconds, which may become a bottleneck for large-scale production lines. In addition, the installation and debugging of line scan cameras are relatively complex and have high environmental requirements, which also increases the cost of maintenance and operation.

[0003] (2) Traditional inspection equipment also relies on high-brightness combination line light sources to ensure image uniformity and clarity. High-brightness combination line light sources provide uniform lighting conditions throughout the entire inspection area through a combination of multiple high-brightness LED light strips and glass rods, ensuring image quality and consistency. However, the cost of this light source is relatively high, further increasing the overall investment in the equipment. In addition, the installation and maintenance of high-brightness combination line light sources are also relatively complicated, requiring precise optical adjustments to ensure uniform distribution of light, which places high demands on the operator's technical level.

[0004] (3) Although the combination of high-resolution line scan cameras and high-brightness combined line light sources can provide high-precision inspection results, there are significant limitations in terms of cost and space. First, high-resolution cameras and high-brightness combined line light sources are expensive, which increases the initial investment in the equipment. Second, due to the long working distance of line scan cameras (25 to 30 cm), the overall size of the inspection equipment is large and occupies more installation space. This is particularly disadvantageous in the transformation of old model production equipment, because these equipment usually already occupy limited space and it is difficult to accommodate additional large and old inspection equipment. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a solar cell detection method and system, which can improve the economy, efficiency and flexibility of detection while ensuring the detection accuracy.

[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is: A solar cell detection method comprises the following steps: Use strip light source and array camera to collect multiple frames of solar cell images; splicing the multiple frames of solar cell images to obtain a solar cell image to be inspected; Performing gamma correction on the solar cell image to be inspected to obtain a corrected solar cell image to be inspected; Defect detection is performed on the corrected image of the solar cell to be detected to obtain a detection result.

[0007] In order to solve the above technical problems, another technical solution adopted by the present invention is: A solar cell detection system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented: Use strip light source and array camera to collect multiple frames of solar cell images; splicing the multiple frames of solar cell images to obtain a solar cell image to be inspected; Performing gamma correction on the solar cell image to be inspected to obtain a corrected solar cell image to be inspected; Defect detection is performed on the corrected image of the solar cell to be detected to obtain a detection result.

[0008] The beneficial effects of the present invention are as follows: a plurality of frames of solar cell images are collected by using a strip light source and an array camera, the plurality of frames of solar cell images are spliced ​​to obtain an image of a solar cell to be detected, a gamma correction is performed on the image of the solar cell to be detected to obtain a corrected image of the solar cell to be detected, a defect detection is performed on the corrected image of the solar cell to be detected to obtain a detection result, and the solar cell images are collected by using the strip light source and the array camera, which greatly reduces the cost of the camera and the light source and saves the installation space of the equipment. Moreover, since the resolution of the array camera is relatively low, the image detection time can be shortened and the detection efficiency can be improved. At the same time, the image is optimized by using image splicing and gamma correction, which solves the problem of uneven brightness caused by the strip light source and improves the uniformity of the overall brightness of the image. It can ensure a high detection rate and a low false positive rate, and ensure the accuracy of the detection result, thereby improving the economy, efficiency and flexibility of the detection while ensuring the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flow chart of the steps of a solar cell detection method according to an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a solar cell detection system according to an embodiment of the present invention; Figure 3 Schematic diagram of an image of a solar cell to be inspected in a method for inspecting solar cells according to an embodiment of the present invention; Figure 4 It is a schematic diagram of a corrected image of a solar cell to be inspected in a solar cell inspection method according to an embodiment of the present invention; Figure 5 It is a schematic diagram of the image rotation process in the solar cell detection method according to an embodiment of the present invention; Figure 6 A schematic diagram of a rectangular detection box in a solar cell detection method according to an embodiment of the present invention; Figure 7 Schematic diagram of the intersection area of ​​the first area and the second area in the solar cell detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0010] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0011] Please refer to Figure 1 , a solar cell detection method, comprising the steps of: Use strip light source and array camera to collect multiple frames of solar cell images; splicing the multiple frames of solar cell images to obtain a solar cell image to be inspected; Performing gamma correction on the solar cell image to be inspected to obtain a corrected solar cell image to be inspected; Defect detection is performed on the corrected image of the solar cell to be detected to obtain a detection result.

[0012] From the above description, it can be seen that the beneficial effects of the present invention are: using a strip light source and an area array camera to collect multiple frames of solar cell images, splicing the multiple frames of solar cell images to obtain a solar cell image to be detected, performing gamma correction on the solar cell image to be detected to obtain a corrected solar cell image to be detected, performing defect detection on the corrected solar cell image to be detected to obtain a detection result, and collecting solar cell images through a strip light source and an area array camera greatly reduces the cost of the camera and the light source, and saves the installation space of the equipment. Since the resolution of the area array camera is relatively low, the image detection time can be shortened and the detection efficiency can be improved. At the same time, image stitching and gamma correction are used to optimize the image, which solves the problem of uneven brightness caused by the strip light source, improves the uniformity of the overall brightness of the image, and can ensure a high detection rate and a low misjudgment rate, thereby ensuring the accuracy of the detection results, thereby improving the economy, efficiency and flexibility of the detection while ensuring the detection accuracy.

[0013] Furthermore, the defect detection is performed on the corrected image of the solar cell to be detected to obtain the detection result, which includes: Performing a blank cell detection on the corrected image of the solar cell to be detected to obtain a blank cell detection result; If the blank sheet detection result is that the corrected solar cell image to be detected is a blank sheet, stopping the detection; If the blank cell detection result is that the corrected solar cell image to be detected is not a blank cell, a solar cell area is acquired from the corrected solar cell image to be detected, and defect detection is performed on the solar cell area to obtain a detection result.

[0014] From the above description, it can be seen that before defect detection, blank film detection is performed first to ensure that solar cells exist in the image for defect detection, thereby ensuring the effectiveness of solar cell detection.

[0015] Further, performing blank detection on the corrected solar cell image to be detected to obtain a blank detection result includes: Selecting pixels whose grayscale values ​​are greater than a first preset value from the corrected solar cell image to be detected as potential solar cell regions; Calculate the area of ​​the largest connected region in the potential solar cell region; Determining the overall image area of ​​the corrected solar cell image to be inspected; Determine whether the ratio of the area of ​​the maximum connected region to the area of ​​the whole image reaches a preset ratio. If so, determine that the empty film detection result is that the corrected solar cell image to be detected is not an empty film, and mark the potential solar cell area as a solar cell area. If not, determine that the empty film detection result is that the corrected solar cell image to be detected is an empty film.

[0016] From the above description, it can be seen that it is more reliable to select pixels with grayscale values ​​greater than the first preset value as potential solar cell areas, calculate the area of ​​the largest connected area, and determine whether the solar cell image to be detected is a blank according to the ratio of the area of ​​the largest connected area to the area of ​​the entire image.

[0017] Furthermore, the defect detection is performed on the solar cell region to obtain the detection result, which includes: Performing circumscribed rectangle fitting on the solar cell region to obtain a circumscribed rectangle; Draw two vertical lines on the upper side or the lower side of the circumscribed rectangle to obtain two intersection points; Calculating the angle between the two intersection points and the horizontal plane; Rotating the corrected image of the solar cell to be inspected according to the angle until the angle between the two intersection points and the horizontal plane is zero, thereby obtaining a rotated image of the solar cell to be inspected; Splitting the color channels of the rotated solar cell image to be inspected to obtain an RGB format image, an HSV format image, and a LAB format image; re-determining the solar cell area from the rotated solar cell image to be inspected; Performing large fragment detection on the solar cell area to obtain a first detection result; Performing a debris print detection on the solar cell region based on the HSV format image and the RGB format image to obtain a second detection result; Performing a pancake detection on the solar cell region based on the RGB format image to obtain a third detection result; A rainbow detection is performed on the solar cell region based on the LAB format image and the HSV format image to obtain a fourth detection result.

[0018] From the above description, it can be seen that the circumscribed rectangle of the solar cell area is fitted to obtain the circumscribed rectangle, the angle is calculated based on the circumscribed rectangle, and the image of the solar cell to be inspected is rotated and corrected according to the angle until the angle between the two intersection points and the horizontal plane is zero. Then defect detection is performed. This can speed up the detection speed and improve the efficiency and reliability of subsequent defect detection. During defect detection, large fragment detection, debris print detection, large pancake detection and rainbow fragment detection are performed, which can accurately detect defects in solar cells and improve the comprehensiveness of defect detection.

[0019] Further, the large fragment detection is performed on the solar cell area to obtain a first detection result, which includes: Performing circumscribed rectangle fitting on the solar cell region to obtain a circumscribed rectangle; Performing difference calculation on the circumscribed rectangle and the solar cell region to obtain a difference region; Determining the width and height of the difference region; It is determined whether the width is greater than a first preset threshold and whether the height is greater than a second preset threshold. If both are true, it is determined that the first detection result is that large fragments exist; otherwise, it is determined that the first detection result is that large fragments do not exist.

[0020] From the above description, it can be seen that performing difference calculation on the circumscribed rectangle and the solar cell area to obtain the difference area, and judging whether there is a large fragment according to the width and height of the difference area is simple and effective.

[0021] Further, the performing debris print detection on the solar cell area based on the HSV format image and the RGB format image to obtain a second detection result includes: Select an S channel image from the HSV format image; Perform edge detection on the S channel image using the Laplace operator to obtain the outline of the debris print; Expanding the debris print outline by a preset pixel to obtain a debris print outline area; Selecting pixels whose grayscale values ​​are less than a second preset value in the S channel image as potential debris print areas; Determine the average grayscale value of the potential debris print area in the R channel image in the RGB format image; Determine whether the average gray value is greater than a third preset threshold value, and if so, determine the potential debris print area as an important potential debris print area, and perform intersection calculation on the debris print contour area and the important potential debris print area to obtain an intersection area; It is determined whether the area of ​​the intersection region is greater than zero. If so, it is determined that the second detection result is that there is a debris mark. If not, it is determined that the second detection result is that there is no debris mark.

[0022] From the above description, it can be seen that the S channel image is firstly subjected to edge detection using the Laplace operator to obtain the outline of the debris print, pixels in the S channel image whose grayscale values ​​are less than the second preset value are selected as potential debris print areas, and the average grayscale value of the potential debris print area in the R channel image is determined in the RGB format image. When the average grayscale value is greater than the third preset threshold, the potential debris print area is determined as an important potential debris print area, and the intersection of the debris print outline area and the important potential debris print area is calculated to obtain the intersection area. When there is an intersection area, it is considered that debris prints exist, thereby accurately detecting the debris prints in the solar cell.

[0023] Further, the performing pancake detection on the solar cell area based on the RGB format image to obtain a third detection result includes: Determine a rectangular detection box at four corners of the solar cell area respectively; Selecting a B channel image from the RGB format image; Calculate the grayscale difference between the maximum grayscale and the minimum grayscale of the area corresponding to each rectangular detection box in the B channel image; Determine whether there are at least three areas whose grayscale differences are greater than a fourth preset threshold. If so, determine that the third detection result is that a large pancake exists; if not, determine that the third detection result is that no large pancake exists.

[0024] From the above description, it can be seen that a B channel image is selected in the RGB image, and the grayscale difference between the maximum grayscale and the minimum grayscale of the area corresponding to each rectangular detection box in the B channel image is calculated. It is judged whether the grayscale difference is greater than the fourth preset threshold, so as to determine whether there is a large pancake, thereby improving the reliability of defect detection.

[0025] Further, the performing rainbow detection on the solar cell region based on the LAB format image and the HSV format image to obtain a fourth detection result includes: Selecting a B channel image from the LAB format image; Selecting pixels whose grayscale values ​​are greater than a third preset value from the B channel image as the first region; Selecting an H channel image from the HSV format image; Selecting pixels whose grayscale values ​​are less than a fourth preset value from the H channel image as the second area; The intersection of the first area and the second area is calculated to obtain an intersection area, and it is determined whether the area of ​​the intersection area is greater than zero. If so, it is determined that the fourth detection result is that a rainbow patch exists; if not, it is determined that the fourth detection result is that a rainbow patch does not exist.

[0026] From the above description, it can be seen that since the production of rainbow flakes is caused by burning in the process, the yellow-brown color is used as the feature for judgment. The yellow-brown defect has a relatively large grayscale value in the B channel of the LAB color space and has obvious characteristics. Therefore, the B channel image is selected from the LAB form image, and pixels with grayscale values ​​greater than the third preset value are selected as the first area. The colors in the first area are all yellow parts. Then, the H channel image is selected from the HSV form image, and pixels with grayscale values ​​less than the fourth preset value are selected as the second area. The colors in the second area are all brown parts. Then the intersection of the two areas is calculated, and the intersection area is yellow-brown in color. If there is an intersection area, that is, its area is greater than zero, it means that a rainbow flake exists, thereby achieving accurate and reasonable rainbow flake defect detection.

[0027] Furthermore, performing gamma correction on the solar cell image to be inspected to obtain a corrected solar cell image to be inspected includes: Determining a gamma value according to the brightness distribution of the solar cell image to be detected; Gamma correction is performed on the solar cell image to be inspected according to the gamma value to obtain a corrected solar cell image to be inspected.

[0028] From the above description, it can be seen that the appropriate gamma value is determined according to the brightness distribution of the solar cell image to be inspected, and then the gamma correction is performed on the solar cell image to be inspected according to the gamma value. The overall brightness of the image is significantly improved, making the brightness of different areas more balanced, and the brightness of the edge area is more consistent with the middle area, thereby improving the image quality and the accuracy and reliability of subsequent defect detection.

[0029] Please refer to Figure 2 Another embodiment of the present invention provides a solar cell detection system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of the above-mentioned solar cell detection method when executing the computer program.

[0030] The above-mentioned solar cell detection method and system of the present invention can be applied to solar cell detection scenarios, and are described below through specific implementation methods: Please refer to Figure 1 , Figure 3-Figure 7 ,in Figure 3 , Figure 4 , Figure 6 and Figure 7 The images collected in actual applications should be dark in color. Figure 5 As shown in, the first embodiment of the present invention is: A solar cell detection method comprises the following steps: S1. Use a strip light source and an array camera to collect multiple frames of solar cell images.

[0031] The strip light source illuminates the solar cell at a low angle (such as 15-30 degrees) and a close distance (such as 95mm) on the side of the solar cell, replacing the traditional high-brightness combined line light source, reducing the detection cost, and providing sufficient brightness in the middle area of ​​the solar cell to meet basic detection needs. The area array camera has a fisheye lens, and simulates the image acquisition logic of the line scan camera under the solar cell, collects solar cell images frame by frame, and obtains multiple frames of solar cell images. The horizontal resolution of each solar cell image is 1300 pixels and the vertical resolution is 5-10 lines of pixels, ensuring that the image area collected each time is small enough for fast processing and splicing, efficiently capturing the complete image of the entire solar cell without missing any details, and improving the response speed of the system due to the reduction in data volume.

[0032] S2. splicing the multiple frames of solar cell images to obtain a solar cell image to be inspected.

[0033] The resolution of the solar cell image to be inspected is 1300×800 pixels, and the background is completely black, which ensures the clarity and contrast of the image.

[0034] S3, performing gamma correction on the solar cell image to be detected to obtain a corrected solar cell image to be detected, such as Figure 5 As shown, specifically including S3.1-S3.2: The brightness of the complete image after stitching is higher in the middle area, which basically meets the detection requirements, but the brightness of the edge area is darker, showing obvious brightness unevenness, such as Figure 3 As shown in the figure. This brightness difference not only affects the overall quality of the image, but may also have an adverse effect on the subsequent defect detection accuracy. Therefore, further brightness optimization is required to ensure the uniformity and consistency of the image, as described below.

[0035] S3.1. Determine a gamma value according to the brightness distribution of the solar cell image to be inspected.

[0036] S3.2. Perform gamma correction on the solar cell image to be inspected according to the gamma value to obtain a corrected solar cell image to be inspected.

[0037] Gamma correction makes the brightness of different areas more balanced by nonlinearly adjusting the brightness of the image. The overall brightness of the image is significantly improved, and the brightness of the edge area is more consistent with the middle area. Figure 4As shown, this not only improves the quality of the image, but also enhances the accuracy and reliability of defect detection.

[0038] S4, performing defect detection on the corrected image of the solar cell to be detected to obtain a detection result, specifically including S4.1-S4.3: S4.1, perform blank detection on the corrected solar cell image to be detected, and obtain a blank detection result, such as Figure 5 As shown, specifically including S4.1.1-S4.1.4: S4.1.1. Select pixels whose grayscale values ​​are greater than a first preset value from the corrected solar cell image to be inspected as potential solar cell regions.

[0039] In an optional implementation, the first preset value is 70.

[0040] Specifically, pixels with grayscale values ​​greater than 70 are selected from the corrected solar cell image to be detected as potential solar cell regions, which can effectively distinguish the solar cell from the background.

[0041] S4.1.2. Calculate the area of ​​the largest connected region in the potential solar cell region.

[0042] S4.1.3. Determine the overall image area of ​​the corrected solar cell image to be inspected.

[0043] S4.1.4. Determine whether the ratio of the area of ​​the maximum connected region to the area of ​​the whole image reaches a preset ratio. If so, determine that the empty film detection result is that the corrected solar cell image to be detected is not an empty film, and mark the potential solar cell area as a solar cell area (SolarRegion). If not, determine that the empty film detection result is that the corrected solar cell image to be detected is an empty film.

[0044] In an optional implementation, the preset ratio is 70%.

[0045] S4.2. If the blank cell detection result is that the corrected solar cell image to be detected is a blank cell, then the detection is stopped.

[0046] S4.3. If the blank cell detection result is that the corrected solar cell image to be detected is not a blank cell, a solar cell region is obtained from the corrected solar cell image to be detected, and defect detection is performed on the solar cell region to obtain a detection result such as Figure 5 As shown, specifically including S4.3.1-S4.3.11: S4.3.1. If the blank cell detection result is that the corrected solar cell image to be detected is not a blank cell, then obtaining a solar cell region from the corrected solar cell image to be detected.

[0047] S4.3.2. Fit a circumscribed rectangle to the solar cell region to obtain a circumscribed rectangle.

[0048] S4.3.3. Draw two vertical lines on the upper or lower side of the circumscribed rectangle to obtain two intersection points.

[0049] S4.3.4. Calculate the angle between the two intersection points and the horizontal plane.

[0050] S4.3.5. Rotate the corrected image of the solar cell to be inspected according to the angle until the angle between the two intersection points and the horizontal plane is zero, thereby obtaining a rotated image of the solar cell to be inspected.

[0051] For example, if the angle is 30 degrees, the corrected image of the solar cell to be inspected is rotated 30 degrees until the angle becomes 0, thereby obtaining the rotated image of the solar cell to be inspected.

[0052] S4.3.6. Split the color channels of the rotated image of the solar cell to be inspected to obtain an RGB format image, an HSV format image, and a LAB format image.

[0053] S4.3.7. Re-determine the solar cell area from the rotated solar cell image to be inspected.

[0054] Specifically, pixels with grayscale values ​​greater than the first preset value are selected from the rotated solar cell image to be detected as potential solar cell areas; and the potential solar cell areas are marked as solar cell areas.

[0055] S4.3.8. Perform large fragment detection on the solar cell area to obtain the first detection result, which specifically includes S4.3.8.1-S4.3.8.4: S4.3.8.1. Fit a circumscribed rectangle to the solar cell region to obtain a circumscribed rectangle.

[0056] S4.3.8.2. Perform a difference calculation on the circumscribed rectangle and the solar cell area to obtain a difference area.

[0057] S4.3.8.3. Determine the width and height of the difference region.

[0058] S4.3.8.4. Determine whether the width is greater than a first preset threshold, and whether the height is greater than a second preset threshold. If both are true, determine that the first detection result is that large fragments exist; otherwise, determine that the first detection result is that large fragments do not exist.

[0059] The first preset threshold and the second preset threshold can be flexibly set according to actual conditions.

[0060] S4.3.9, based on the HSV format image and the RGB format image, the solar cell area is subjected to a debris print detection to obtain a second detection result, specifically including S4.3.9.1-S4.3.9.7: S4.3.9.1. Select an S channel image from the HSV format image.

[0061] S4.3.9.2. Use the Laplacian operator to perform edge detection on the S channel image to obtain the debris print outline.

[0062] S4.3.9.3. Dilate the debris print outline by a preset number of pixels to obtain a debris print outline area.

[0063] In an optional implementation, the preset pixels are 20 pixels.

[0064] S4.3.9.4. Select pixels whose grayscale values ​​are less than a second preset value in the S channel image as potential debris print areas.

[0065] In an optional implementation, the second preset value is 80.

[0066] S4.3.9.5. Determine the average grayscale value of the potential debris print area in the R channel image in the RGB format image.

[0067] S4.3.9.6. Determine whether the average grayscale value is greater than a third preset threshold value. If not, determine that the second detection result is that there is no debris print. If so, determine the potential debris print area as an important potential debris print area, and calculate the intersection of the debris print contour area and the important potential debris print area to obtain the intersection area.

[0068] The third preset threshold can be flexibly set according to actual conditions.

[0069] S4.3.9.7. Determine whether the area of ​​the intersection region is greater than zero. If so, determine that the second detection result is that there is a debris mark. If not, determine that the second detection result is that there is no debris mark.

[0070] S4.3.10. Perform a pancake detection on the solar cell area based on the RGB image to obtain a third detection result, specifically including S4.3.10.1-S4.3.10.4: S4.3.10.1. Determine a rectangular detection box at each of the four corners of the solar cell area, such as Figure 6 shown.

[0071] S4.3.10.2. Select a B channel image from the RGB format image.

[0072] S4.3.10.3. Calculate the grayscale difference between the maximum grayscale and the minimum grayscale of the area corresponding to each rectangular detection box in the B channel image.

[0073] S4.3.10.4. Determine whether there are at least three regions whose grayscale differences are greater than a fourth preset threshold value. If so, determine that the third detection result is that a large pancake exists. If not, determine that the third detection result is that no large pancake exists.

[0074] The fourth preset threshold can be flexibly set according to actual conditions.

[0075] That is, when the grayscale difference of the areas corresponding to three or more rectangular detection boxes is greater than the fourth preset threshold, it can be determined that a large pancake exists.

[0076] S4.3.11. Perform rainbow detection on the solar cell area based on the LAB format image and the HSV format image to obtain a fourth detection result, specifically including S4.3.11.1-S4.3.11.5: S4.3.11.1. Select a B channel image from the LAB format image.

[0077] S4.3.11.2. Select pixels whose grayscale values ​​are greater than a third preset value from the B channel image as the first region.

[0078] In an optional implementation, the third preset value is 170.

[0079] S4.3.11.3. Select an H channel image from the HSV format image.

[0080] S4.3.11.4. Select pixels whose grayscale values ​​are less than a fourth preset value from the H channel image as the second area.

[0081] In an optional implementation, the fourth preset value is 50.

[0082] S4.3.11.5. Calculate the intersection of the first region and the second region to obtain an intersection region, such as Figure 7 As shown, it is determined whether the area of ​​the intersection region is greater than zero. If so, it is determined that the fourth detection result is that a rainbow patch exists. If not, it is determined that the fourth detection result is that a rainbow patch does not exist.

[0083] The present invention replaces the expensive 5-megapixel line scan camera by using the common fisheye lens and 1.3-megapixel area array camera on the market, greatly reducing the cost of the camera and lens. The strip light source is used instead of the traditional high-brightness combined line light source, further reducing the cost of the light source and simplifying the design and maintenance of the light source. By simulating the image acquisition logic of the line scan camera, only 5-10 rows of pixels are captured per frame, which greatly reduces the image acquisition time. At the same time, due to the use of a low-resolution camera, the algorithm detection time is shortened from the original 200 milliseconds to 50 milliseconds, so that the total detection time is shortened to less than 100 milliseconds, significantly improving the detection efficiency. In addition, the camera's working distance is reduced, which significantly reduces the installation space of the detection equipment. It is particularly suitable for the transformation of old model production equipment and saves valuable production space.

[0084] The introduction of image stitching technology and gamma correction technology can ensure a high detection rate and a low false positive rate despite the use of low-resolution cameras, thereby ensuring the accuracy of the detection results. In particular, gamma correction can solve the problem of uneven brightness caused by strip light sources, improve the overall brightness uniformity of the image, and further improve the reliability and accuracy of detection.

[0085] Please refer to Figure 2 , Embodiment 2 of the present invention is: A solar cell detection system comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step of the solar cell detection method in the first embodiment is implemented.

[0086] In summary, the present invention provides a solar cell inspection method and system, which utilizes a strip light source and an array camera to collect multiple frames of solar cell images, splices the multiple frames of solar cell images to obtain an image of a solar cell to be inspected, performs gamma correction on the image of the solar cell to be inspected to obtain a corrected image of the solar cell to be inspected, performs defect inspection on the corrected image of the solar cell to be inspected, and obtains an inspection result. By collecting the solar cell image through the strip light source and the array camera, the cost of the camera and the light source is greatly reduced, and the installation space of the equipment is saved. Since the resolution of the array camera is relatively low, the image inspection time can be shortened and the inspection efficiency can be improved. At the same time, image splicing and gamma correction are used to optimize the image, and the brightness unevenness caused by the strip light source is solved. Uniformity problem, improves the uniformity of the overall brightness of the image, can ensure a high detection rate and a low false positive rate, ensure the accuracy of the detection results, thereby improving the economy, efficiency and flexibility of the detection while ensuring the detection accuracy; in addition, the solar cell area is fitted with a circumscribed rectangle to obtain a circumscribed rectangle, the angle is calculated based on the circumscribed rectangle, and the image of the solar cell to be detected is rotated and corrected according to the angle until the angle between the two intersection points and the horizontal plane is zero, and then defect detection is performed, which can speed up the detection speed and improve the efficiency and reliability of subsequent defect detection. During defect detection, large fragment detection, debris print detection, large pancake detection and rainbow fragment detection are performed, which can accurately detect the defects of the solar cell and improve the comprehensiveness of defect detection.

[0087] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A solar cell detection method, characterized in that: Includes steps: Use strip light source and array camera to collect multiple frames of solar cell images; splicing the multiple frames of solar cell images to obtain a solar cell image to be inspected; Performing gamma correction on the solar cell image to be inspected to obtain a corrected solar cell image to be inspected; Defect detection is performed on the corrected image of the solar cell to be detected to obtain a detection result.

2. A solar cell detection method according to claim 1, characterized in that: The defect detection is performed on the corrected image of the solar cell to be detected to obtain the detection result, which includes: Performing a blank cell detection on the corrected image of the solar cell to be detected to obtain a blank cell detection result; If the blank sheet detection result is that the corrected solar cell image to be detected is a blank sheet, stopping the detection; If the blank cell detection result is that the corrected solar cell image to be detected is not a blank cell, a solar cell area is acquired from the corrected solar cell image to be detected, and defect detection is performed on the solar cell area to obtain a detection result.

3. A solar cell detection method according to claim 2, characterized in that: The performing blank detection on the corrected solar cell image to be detected to obtain a blank detection result comprises: Selecting pixels whose grayscale values ​​are greater than a first preset value from the corrected solar cell image to be detected as potential solar cell regions; Calculate the area of ​​the largest connected region in the potential solar cell region; Determining the overall image area of ​​the corrected solar cell image to be inspected; Determine whether the ratio of the area of ​​the maximum connected region to the area of ​​the whole image reaches a preset ratio. If so, determine that the empty film detection result is that the corrected solar cell image to be detected is not an empty film, and mark the potential solar cell area as a solar cell area. If not, determine that the empty film detection result is that the corrected solar cell image to be detected is an empty film.

4. A solar cell detection method according to claim 2, characterized in that: The performing defect detection on the solar cell region to obtain the detection result comprises: Performing circumscribed rectangle fitting on the solar cell region to obtain a circumscribed rectangle; Draw two vertical lines on the upper side or the lower side of the circumscribed rectangle to obtain two intersection points; Calculating the angle between the two intersection points and the horizontal plane; Rotating the corrected image of the solar cell to be inspected according to the angle until the angle between the two intersection points and the horizontal plane is zero, thereby obtaining a rotated image of the solar cell to be inspected; Splitting the color channels of the rotated solar cell image to be inspected to obtain an RGB format image, an HSV format image, and a LAB format image; re-determining the solar cell area from the rotated solar cell image to be inspected; Performing large fragment detection on the solar cell area to obtain a first detection result; Performing a debris print detection on the solar cell region based on the HSV format image and the RGB format image to obtain a second detection result; Performing a pancake detection on the solar cell region based on the RGB format image to obtain a third detection result; A rainbow detection is performed on the solar cell region based on the LAB format image and the HSV format image to obtain a fourth detection result.

5. A solar cell detection method according to claim 4, characterized in that: The large-fragment detection of the solar cell region is performed to obtain a first detection result, comprising: Performing circumscribed rectangle fitting on the solar cell region to obtain a circumscribed rectangle; Performing difference calculation on the circumscribed rectangle and the solar cell region to obtain a difference region; Determining the width and height of the difference region; It is determined whether the width is greater than a first preset threshold and whether the height is greater than a second preset threshold. If both are true, it is determined that the first detection result is that large fragments exist; otherwise, it is determined that the first detection result is that large fragments do not exist.

6. A solar cell detection method according to claim 4, characterized in that: The performing debris print detection on the solar cell region based on the HSV format image and the RGB format image to obtain a second detection result includes: Select an S channel image from the HSV format image; Perform edge detection on the S channel image using the Laplace operator to obtain the outline of the debris print; Expanding the debris print outline by a preset pixel to obtain a debris print outline area; Selecting pixels whose grayscale values ​​are less than a second preset value in the S channel image as potential debris print areas; Determine the average grayscale value of the potential debris print area in the R channel image in the RGB format image; Determine whether the average gray value is greater than a third preset threshold value, and if so, determine the potential debris print area as an important potential debris print area, and perform intersection calculation on the debris print contour area and the important potential debris print area to obtain an intersection area; It is determined whether the area of ​​the intersection region is greater than zero. If so, it is determined that the second detection result is that there is a debris mark. If not, it is determined that the second detection result is that there is no debris mark.

7. A solar cell detection method according to claim 4, characterized in that: The performing pancake detection on the solar cell region based on the RGB format image to obtain a third detection result includes: Determine a rectangular detection box at four corners of the solar cell area respectively; Selecting a B channel image from the RGB format image; Calculate the grayscale difference between the maximum grayscale and the minimum grayscale of the area corresponding to each rectangular detection box in the B channel image; Determine whether there are at least three areas whose grayscale differences are greater than a fourth preset threshold. If so, determine that the third detection result is that a large pancake exists; if not, determine that the third detection result is that no large pancake exists.

8. A solar cell detection method according to claim 4, characterized in that: The performing rainbow sheet detection on the solar cell area based on the LAB form image and the HSV form image to obtain a fourth detection result includes: Selecting a B channel image from the LAB format image; Selecting pixels whose grayscale values ​​are greater than a third preset value from the B channel image as the first region; Selecting an H channel image from the HSV format image; Selecting pixels whose grayscale values ​​are less than a fourth preset value from the H channel image as the second area; The intersection of the first area and the second area is calculated to obtain an intersection area, and it is determined whether the area of ​​the intersection area is greater than zero. If so, it is determined that the fourth detection result is that a rainbow patch exists; if not, it is determined that the fourth detection result is that a rainbow patch does not exist.

9. A solar cell detection method according to claim 1, characterized in that: The performing gamma correction on the to-be-detected solar cell image to obtain the corrected to-be-detected solar cell image comprises: Determining a gamma value according to the brightness distribution of the solar cell image to be detected; Gamma correction is performed on the solar cell image to be inspected according to the gamma value to obtain a corrected solar cell image to be inspected.

10. A solar cell detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, each step of the solar cell detection method according to any one of claims 1 to 9 is implemented.

Images