Development of visual inspection system for low-reflective material utilizing of a string shadow

Visual inspection of products in manufacturing plant is usually conducted by workers.However, that inspection process by workers occasionally happen oversight due to careless check, so that automatic visual inspection system is needed from manufacturer.Products made of low-reflective material are difficult to inspect by light reflection because contours of defect areas are got blurred.To solve this problem, we propose to inspect products utilizing of a string shadow.Form our research, a string shadow was different between contours of defect area and its of non-defect.On the other hand, former research did not success to detect defect area due to determination algorithm for defect.In this paper, we apply linear approximation of a string shadow to defect detection.From the experimental results, we succeed in detecting defects.