Abstract: Due to the image noise and the instability of the acquisition system, there is often contour noise interference in the conventional differential detection algorithm. To address this situation, an improved contour suppression differential defect detection algorithm is proposed in conjunction with the round PIN defect detection task. The HSV color space was introduced and quantitative defects were detected by using HS two-channel color segmentation method and extracting quantitative features. A chromaticity-free luminance map was constructed by extracting V-channel images. A multi-scale contour suppression strategy was proposed to overcome the problem of excessive contour noise of traditional differential methods and complete the detection of qualitative defects. The experimental results show that the algorithm proposed in this paper achieves 100% detection accuracy for quantitative defects and 92.2% precision for qualitative defects of circular PINs, with an average detection efficiency of 256ms/root. The algorithm can effectively suppress the interference brought by contour noise, while enhancing the prominence of defects and obtaining good defect detection results.