Visual Inspection for Circular Objects Based on Global Symmetry

The research is aimed to accomplish an automatic visual inspection system for circular objects using cameras and image processing technique. The main concern of the paper is on the pattern analysis of circular objects based on image analysis. First, the authors propose a computational method using global symmetry to locate objects before the further inspection process. It aims at designing a symmetry measure based on distance weight, phase weight and intensity weight. Such measure can be used to locate centers of circles, even for those with weak contrast under uncertain complex backgrounds. Then based on the measure, similarities of arbitrary circular objects are given and tested. The experimental results for the proposed approach are promising.