Bearing is a basic work-piece in machinery devices, and surface quality of steel ball is the main factor which affects the precision and longevity of bearing. Currently defects of steel ball are detected manually in industry. It is inefficiency and of high probability of misidentification. In order to assure the stability of steel ball quality this paper put forward an autodetection method based on vision technique to detect surface defects of steel ball. Firstly we designed an approach to fully expand the surface of steel ball according to the requirement of image detection. Then we made up a corresponding device to accomplish to designed approach and developed a platform system for image detection. Finally we carried a proving detection in some kind of defects of steel ball. The result of test shows that the method can be put into use to detect the general defects detection of steel ball.

No abstract