Defect Detection Based on Fusion of Phase Congruency and Shear Wavelet

A new defect detection algorithm for patch components based on fusing phase congruency and shear wavelet is proposed. Compared to the gradient-based defect detection, the phase congruency (PC) is insensitive to variations in illumination, contrast, and noise. In this paper, we proposed an improved phase congruency method, which improves the sensitivity to defects and removes the noise interference at the price of the blurring to defect's contours. To solve this problem, we also proposed a fusing strategy to combine phase congruency and shear wavelet, which can not only fit the defect's contour precisely but also remove noise better. To evaluate the performance of the proposed method, we design a variety of experiments on patch components. Finally, the results show our method has excellent performance for the detection of patch components.