Moving Object Detecting System With Phase Discrepancy

This paper proposes an efficient moving object detecting system that detects moving objects in dynamic scene. The system consists of three parts: motion saliency calculation, moving area extraction and bounding box generation. We further analyze the the phase discrepancy algorithm and use it to get the motion saliency map from adjacent images. We use Canny-like salient area extraction algorithm to extract moving segments from motion saliency map. We then use graph based image segmentation algorithm to extend salient; areas to bounding boxes. Computer simulations are given to demonstrate the high performance in detecting moving objects.