Skip to Main Content
The surface defects in copper strips severely affect the quality of copper. So detecting the surface defects in copper strip has great significance to improve the quality. This paper presents a copper strip surface inspection based on computer vision, which uses modularized frame of hardware and the software of image processing. The paper adopts a self-adaptive weight averaging filtering method to preprocess image, and uses the moment invariants to pick the characters of typical defects which eigenvector is identified with the RBF neural networks. Experiments show that the real-time method can effectively detect the copper strip surface defects in the production line.