Image segmentation techniques for real-time coverage measurement in shot peening processes

Abstract

Shot peening is the process of treating metallic surfaces with a regulated blast of shots to increase material strength and durability. Determining the coverage level of the shots is an important parameter in assessment of the quality of treatment. Traditionally, manual coverage measurement is performed visually which is prone to human error and judgement. Despite the proposal for the use of image segmentation techniques for determining the coverage levels, literature on the topic is not extensively developed. Various relevant image segmentation techniques are investigated in this study. In particular, thresholding, edge detection, watershed segmentation, active contour, and graph cut techniques are investigated and applied to shot peen coverage measurement. The results obtained from each method are discussed and compared against a set of relevant performance criteria.