Vision-Based Surface Defect Detection for Flange Nuts

With the global market’s need to improve the quality of fasteners, we aim to develop a defect detection system with computer vision for flange nut. The tracing of boundary and the K-curvature method is conducted on a testbed to search for defect candidates. First, to detect and recognize the defects, it needs to extract the detail of the captured image. Next, based on the 8-neighbor pattern analysis, the one-pixel-wide edge is obtained to represent the binary profile of the flange nut. Finally, estimating K-curvature is implemented to determine whether the concerned segment is a defect one. The experiment result illustrates the validity of the proposed algorithm, which is suitable for detecting the defects of flange nuts.