Zipper classification and defect detection based on computer vision

At present, the research progress of zipper classification and defect detection based on computer vision is slow in China. Generally, manual method is used to detect the appearance quality of zipper products, which is not only inefficient, but also low reliability. This paper proposes a reliable and accurate method. In the actual industrial production, there are various types, shapes and colors of zippers, and in the actual situation, lighting and other factors have a great impact on the detection effect. In view of the above situation, this paper proposes an adaptive region growing algorithm and template matching algorithm. By region growing algorithm, the background is removed to obtain the complete inner region of the zipper. Then, the deflection angle of the zipper is corrected by the smallest external rectangle of the zipper. Finally, the zipper type is determined by template matching algorithm. For defect detection, firstly, the complete zipper is extracted by morphological method, then the defect location is determined by searching the area of connected domain, and finally the defect is determined by gray value comparison. The experimental results show that the above method has high accuracy and execution efficiency in zipper classification and defect detection.