A Machine Vision System for Bearing Greasing Procedure

Bearing is widely used in many machines, which can reduce the friction between connected components. However, many manufacturers still use human or machine methods to inspect the bearing production, which is inefficient, costly and unreliable, especially for the miniature bearing. In this paper, we propose a machine vision system for bearing greasing procedure. The proposed system uses image processing technology to process digital image captured by a camera and can locate the bearing cage quickly and accurately. Firstly, the bearing is separated from the whole image using the RANSAC least square circle fitting method. Secondly, to facilitate the process algorithm, the bearing area is transformed into a rectangle image. Next, some novel projection methods are involved. Finally, a center map is calculated to get the final greasing location. Experimental results show that the proposed machine vision system has high accuracy and efficiency, and can fully meet the online production requirement.