The Method of Thread Defect Detection Based on Machine Vision

摘要

To resolve the problem of low accuracy and difficulty in detecting threads, this paper proposes a method of detecting based on machine vision. The image of the threaded part is collected online by an industrial CCD camera, and the image is gray scaled, enhanced, binary and de-noised. The de-noising method selects the mathematical morphology operation. Mathematical morphology of corrosion, expansion, open and closed operations effectively preserves thread shape characteristics and edge information of the threads. The algorithm detects the position of the thread defect by locating the defect of the thread using a subtraction operation. The experimental results show that the algorithm has strong ability to identify and detect defects and can effectively identify appearance defects such as no thread, less thread and irregular pitch, which meets the basic design requirements.