Research on Defect Detection Method of Painting Parts Based on Machine Vision

Abstract

The technology is used to study and analyze the appearance defect characteristics of sprayed auto parts. The corresponding recognition method is developed by using MATLAB and Halcon software. Firstly, the causes of background noise and uneven brightness are described from the hardware and theory, and the specific treatment methods are given from the two perspectives of eliminating background noise and balancing illumination. A variety of frequency domain low-pass filters are constructed to eliminate background noise. The image preprocessing algorithm based on homomorphic filter is studied to achieve the expected preprocessing effect. The image segmentation method of local features is adopted, and the binary processing is carried out. Corresponding recognition algorithms are designed for different defect features. The on-line monitoring system of spraying quality is established and field test is carried out to verify the effectiveness of the system and realize the rapid and accurate identification of spraying quality. The single recognition time is 320 milliseconds, and the recognition accuracy is 97%. Compared with manual identification, it is improved by about 40% and meets the needs of industry applications