Segmentation defects in the solar cell module detection system plays a very important role. In this paper, two peaks and maximum between-class variance of solar near-infrared image segmentation and image binarization and binary image obtained defects. Experiments show that the proposed silicon solar cell image segmentation method can achieve fast and accurate image segmentation, get a good binary image. Defects in the solar image feature extraction, and further description and analysis of laying a good foundation.