Defect detection of polycrystalline solar wafers using local binary mean

Polycrystalline solar wafers consist of various crystals and their surfaces have heterogeneous textures. The conventional defect detection methods cannot be applied to their solar wafers. In this paper, we propose a concept of local binary mean and its optimization method for selecting optimal threshold T. The input image is broken down into a set of K patch images. Each patch image is used to calculate its local binary mean. The local binary mean value is used as the discrimination measure for detecting defects. Experimental results show that our proposed method achieves a detection rate of 91~94 %. Compared with related defect detection methods, the proposed method has the advantage of detecting various kinds of low gray-level defects such as micro-cracks, fingerprints, and contaminations simultaneously.