A Corner-Analysis-Based Method for Dimensional Inspection and Defect Detection of Glass Chamfers

Edge chamfering is a critical process in glass manufacturing because it suppresses crack propagation and improves both edge strength and operational safety. As such, it has become an essential step in modern glass production. However, in current automated production lines, chamfer quality inspection still relies heavily on manual visual inspection and contact-based measuring instruments, which are limited by low efficiency, strong subjectivity, and poor compatibility with production-line takt time. To overcome these limitations, this paper proposes a vision-based inspection method for transparent glass chamfers using corner analysis, and a dedicated glass chamfer inspection system is developed for experimental validation. First, machine-vision image processing is applied to enhance chamfer-region features and improve detection robustness. Candidate endpoint generation is then combined with the construction of an endpoint-screening coordinate system to achieve stable localization of chamfer endpoints. In addition, quadrant-based spatial constraints are introduced to further enhance endpoint localization stability, enabling automatic measurement of key geometric parameters, including chamfer length, width, and height. On this basis, multiple defect-discrimination criteria are established by integrating geometric tolerance violations with spatial abnormalities in endpoint distribution, thereby enabling automatic identification of chamfer defects. Experimental results show that the proposed method achieves a mean absolute measurement error of 0.087 mm and an average inspection time of 0.505 s per workpiece. The method can also effectively identify chamfer defects, demonstrating its suitability for online inspection while maintaining both dimensional measurement accuracy and defect detection capability.