Robust Defect Detection Algorithm for Highly Reflective Surfaces Using SIFT Feature Points and Perceptual Hashing

In the manufacturing industry, materials are widely used in automobiles, electronic products and other fields. However, due to its reflective characteristics, traditional visual inspection algorithms are often interfered with by reflected light when detecting such surface defects, resulting in low inspection accuracy. Therefore, this study proposes a highly reflective surface defect detection algorithm based on scale invariant feature transformation (SIFT) feature points and perceptual hash method, aiming at improving the robustness and accuracy of detection. The algorithm extracts key feature points in highly reflective surface images through SIFT, and compares image similarity with perceptual hash algorithm to accurately identify surface defects. Experimental data show that the accuracy of this algorithm reaches 96.5% in defect detection of multiple highly reflective surfaces (such as mirror metal plates), which is significantly improved compared with 83.7% of traditional edge detection algorithms. At the same time, the false detection rate of the algorithm is reduced from 15.3% to 6.8% in the traditional method, and the missed detection rate is reduced from 12.1% to 4.5%. Through a large number of experiments, it is verified that the defect detection algorithm proposed in this study, which combines SIFT feature points with perceptual hash, has significant advantages in highly reflective surface detection, and can effectively overcome reflection interference and ensure detection accuracy and robustness.