The objective of this study is to improve the efficiency of identifying fracture initiation points in tempered glass by utilizing an integrated system that combines an optical system and image processing techniques. Tempered glass enhances its mechanical strength by inducing compressive stress on its surface. Traditionally, the identification of fracture initiation points in tempered glass depends on manual inspection, which is both time-intensive and reliant on the inspector's experience. To address these challenges, we developed a system that incorporates an optical system for uniform illumination using a light guide plate along with a robust image processing workflow. The workflow includes dark subtraction, binary thresholding using Otsu's binarization method, and contour extraction with the Suzuki85 algorithm. Additionally, polygonal approximation of the contours was performed using the Douglas-Peucker algorithm, with an experimentally determined optimal tolerance value of 4.1%. By analyzing fragment characteristics such as the number of vertices, convexity, area, and aspect ratio, the system was able to identify potential initiation fragments based on criteria derived from experienced fractography inspectors. Experimental validation was conducted using ten tempered glass panels, each fractured under controlled impact conditions to ensure consistency. The developed system successfully reduced the number of fragments that inspectors need to investigate to approximately 10% for the group of fragments arranged with the in-plane direction facing upward within the capture screen. This study demonstrates the practical feasibility of integrating optical and image processing technologies to automate and enhance tempered glass fracture analysis. The proposed system has potential for adaptation to other materials and industrial applications.
Taira et al. (Wed,) studied this question.