Analysis of dynamic images in machine vision and its application study in motion control

Recent advances in high performance computing coupled with the decreasing cost of hardware now make machine vision a financially viable inspection option for even small and medium sized firms. Amongst many problems relative to the machine vision applications, the detection and tracking of moving targets is very important in many cases including industrial fields. Our challenge is to develop an effective methodology on analysis of dynamic images able to extract the moving target by using only visual sensing as input and keeping computation time and hardware cost to a minimum, typically with a standard Pentium-based computer and a standard CCD camera. This paper presents a segmentation method combining dynamic segmentation with static segmentation and examines moving object detection based primarily on inter-frame difference method. We applied the methods into the motion control of a familiar work table driving system only based on machine vision. The experimental results demonstrate that in realistic situations, detection of moving targets by using inter-frame difference method is more effective than that by using optical flow method.