Research On Surface Defect Inspection Of Solar Cell Based On Machine Vision

With the exhaustion of non-renewable resources and the increment for energy, solar energy, as a kind of general, equal and ecological energy, has aroused many countries'attention and got development. As its carrier, solar cells have become a focus. The quality of solar cells has become one of main factors which affect the efficiency of solar modules, so it is very important to enhance inspection for solar cells in the production process. Based on machine vision this article does some research on the defects detection for the surface of solar cells.Aiming at strong reflection of solar cell surface and considering the influence of light source, this article has used a white, ring LED lamp.Because of regular, strong textures characteristic of the surface of solar cell, this article uses different methods for different defects. For broken corners, this article adopts image segmentation, morphology and median filtering to get broken corner image. For cracks and broken grids, First of all, the paper scan image sequentially along the direction of grid lines, analyze the features of defects, get grayscale difference curves from which this system can extract the coordinate of grayscale catastrophe points, then, remove the effect of broken corner by using broken corners'coordinates. Finally, to eliminate interference factors this system process defect image with morphology and median filtering.For defect image, this paper has did some work about defect labeling and defect identification by choosing some corresponding feature parameters as the basis of judging defect types. Based on Matlab this article designed a system interface for the surface inspection of solar cells, from which readers can acquire defects image, defects number and statistical data about defects.Experiments proved that this system can detect broken corners, cracks and broken grids at a higher recognition rate and achieve real-time demand basically.