カメラ画像による光沢曲面上の微小キズ検出の試み

抄録

In the manufacturing industry, automation of visual inspection has been achieved with many objects, but there are objects that are difficult to inspect at real cost. For example, in metal precision machined parts, coolant is used during machining, so the target surface is covered with liquid film. Therefore, two extra steps are required. That is, removing the liquid film immediately before the test and covering again with the liquid film immediately after the inspection. Moreover, it is difficult to conduct a uniform automatic inspection for minute scratches on a curved surface. In this research, we examined image inspection methods for precisely detecting minute defects on the glossy surface without completely removing the liquid film by cleaning, for piston pins which are precision metal processed parts. As the results of image processing experiments, all defective parts were detected.