Rupture Disc Detection Method Based on Curve Detectors

Abstract

In the automated production of bursting discs, this paper proposes a method based on a deep learning edge detection model combined with curve detector post-processing, effectively addressing the challenges of defect detection and positioning for circular blanks. The method first uses an edge detection model to extract the preliminary edge information of the workpiece, followed by a skeletonization algorithm and a segmented curvature detection algorithm to refine the extraction of single-pixel edges and calculate curvature segment by segment. By comparing curvature features, abnormal segments are identified to locate defects, while curve fitting techniques in normal regions achieve precise workpiece positioning. Experimental results demonstrate that the proposed algorithm ensures high accuracy while maintaining high computational efficiency, meeting the real-time requirements of automated production and significantly improving production quality and stability.