Robust Crack Defect Detection in Inhomogeneously Textured Surface of Near Infrared Images

Abstract

Robust crack defect detection in solar cells has been challenging because of the inhomogeneously textured surface, low contrast between crack defect and background, the diversity of crack types, and so on. To overcome these challenges, this paper presents a new robust crack defect detection scheme for multicrystalline solar cells. Firstly, a steerable evidence filter is designed to process EL image to obtain the response map, which enhances the contrast between crack and background and provides evidence for the presence of crack defect. Secondly, complete crack extraction from the response map is employed. Finally, the complete crack can be located in the inspection image by the crack skeleton extraction. Experimental results on defective and defect-free EL images show that the proposed scheme is robust, and various cracks can be effectively detected, which outperforms the previous methods.