Polysilicon Slice Dislocation Defects Segmentation and Area Statistics Based on Curve Fitting

The quality of the polysilicon slice affects the efficiency of polysilicon solar cells directly, dislocation defects exist in the polysilicon generally, a large number of dislocation defects have a greater impact on efficiency of solar cells. On the basis of photoluminescence defects detection, this paper proposes a new method that progressive scan image to obtain the grayscale curve of each row, and do curve fitting for each grayscale curve to achieve defect segmentation, by comparing the segmentation results obtained by quadratic curve and Gaussian curve fitting, proves that the quadratic curve fitting can be better for defects segmentation. At last, get the proportion of defective area in total slice area. Experiments show that the method of quadratic curve fitting is efficiency and accuracy for dislocation defects segmentation and counting defects area ratio.