Open AccessOn the detection of defects on smooth free metallic paint surfaces
Author(s) -
Pan Guo,
Pin Lu Cao,
Peng-Fei Zhang,
Shiwei Chen,
Yongrong Yang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1676/1/012166
Subject(s) - fusion , artificial intelligence , computer vision , template matching , computer science , image fusion , fuse (electrical) , pattern recognition (psychology) , image (mathematics) , engineering , philosophy , linguistics , electrical engineering
In view of the difficulty of automatic detection of defects on smooth free metallic paint surfaces, a simple and effective automatic detection system based on deflection principle is proposed in this paper. This system consists of cameras and a set of striped lights. A series of images can be acquired at a certain frequency by camera. And the detection algorithm is divided into three steps. In the first step, the square-difference fusion method and minimum fusion method are used to fuse a group of the standard images to obtain two fusion images, which are used to extract the ROI and partial template. In the second step, partial template is used to perform template matching and correction on the fusion images. In the last step, extraction of mask, extraction of defects, and screening of defects are performed on the corrected fusion image to extract defects. To detect complex metallic paint surfaces of the car, the experimental results demonstrate that the defects detection accuracy can reach 85% and the defect recognition size can reach 0.1mm.