Detection And Classification Technology Of Solar Cell Defects Based On Digital Image Process

Solar cell as a carrier of the solar energy is widely used, the quality of the solar cell is goodor not, will directly affect their life span and the power generation efficiency. In the currentindustrial production process, if relying on the visual inspection method, the products’ qualitycan not have a good guarantee as the human’s uncertainties. Using of the industrial machinevision on the quality inspection of the industrial is more and more popular, because of it with areliable, efficient, accurate and intelligent advantages relatively the artificial visual inspection.To achieve the large-scale production of the products, product testing consistency has anirreplaceable role, not only it greatly reduces the labor costs. Machine vision involves manyaspects, which on the one hand is the target image acquisition and target image preprocessingand recognition, in this paper, the solar cell defect detection has been studied, regarding thecurrent method with a single defect detection type and the test algorithm of less anti-interferenceability in solar cell defect detecting, a new method which can detect various defections incomplicated background of the solar cell pictures is proposed. The main contents can be dividedas follows:(1) Positioning and removing the battery electrodes based on the linear filtering andhorizontal integral projecting; An improved high-pass emphasis filter with adaptable parametersis used to enhance the defects,and then extracting the defect characteristics by an improvedself-adapting threshold binary process.Getting a good experimental results,the accuracyespecially on the black heart defect can get100percent.(2) Improving a local adaptive thresholding binarization process method, and horizontalprojection can be used to get the positions of electrodes, then filling electrodes.It can detect theblack points, broken lines and crack defects fastly and accurately.(3) A method based on the defections contour information which can classify according todefections contour information is used, by calculating their aspect ratio, area ratio and the center,classifing and identifing based on them.The significance of this paper is that a new method which can detect various defections incomplicated background of the solar cell pictures, greatly improving the accuracy of detectionand identification, with a high practical value.