MOI-Based Stratified Crack Detection: A PCA Approach*

Recently, magneto-optical image (MOI) test for crack detection is developing rapidly and becoming a research focus in the nondestructive testing (NDT) field. In this study, an MOI segmentation method, which is based on the principal component analysis (PCA) algorithm, is proposed to extract meaningful characteristics and identify cracks from raw images. With the discovered physical meaning of the PCA of MOIs, where the first principle component is the background and the second is the black domain spots, the crack information is highlighted by separating the background, magnetic domain noises, and some other information into different components. In addition, a fusion strategy is defined to combine the significant components so as to enhance the crack features further. Experimental magnetic optical image dataset of a specimen is utilized to verify the validity of the proposed method. It implies that the proposed method makes the background information more prominent to easily distinguish the crack and improves the ability of the detection system to detect the defects automatically.