Defect Algorithm for Ultra-Thin Flexible Glass Surfaces Based on Parallel Light Scanning

With the rapid development of the manufacturing industry, product quality inspection standards have become increasingly stringent, leading more and more companies to focus on surface defect detection. Ultra-thin flexible glass, as a novel material, is currently inspected manually. In order to achieve automation through machine vision and reduce labor costs, this paper presents a defect detection algorithm for ultra-thin flexible glass surfaces using parallel light scanning and high-precision cameras to capture backlit images of the glass. Common defects are categorized and analyzed, and a limited sample library is established. The algorithm for defect detection on ultra-thin flexible glass surfaces is proposed based on Halcon operators. This method can detect nine common defects, including uneven etching, fitting, leakage, gap, crack, edge bump, concave point, scratch, and convex point, with a detection accuracy of up to 10 micrometers. Experiments show that the algorithm achieves an accuracy rate of 94.4%, a false alarm rate of 2.24%, and a miss detection rate of 3.36%, effectively assisting in detecting defects in seven pieces of glass that were not identified through manual inspection.