Surface defect analysis of rotary metal parts

Surface defect analysis of rotary metal parts is a requirement to research quality and operating state of parts. This paper proposes a new approach and algorithms for surface defect analysis of rotary metal parts. Given the captured surface images as an input, the approach first uses the pose relation of the mechanism and a simple weighted average algorithm to obtain one seamless large image by fast stitching of the measured multi-view images. Then, to perform initial segmentation of surface defects, the possible defects are segmented using the developed segmentation algorithm based on contourlet transform with adaptive thresholds (CTAT) and the improved active contour model (ACM), and the defect parameters can be obtained and evaluated. Experimental results on inspection of the surface defects for a variety of rotary metal parts are reported to show the validity of the proposed methodology.