A new moving object contour detection approach

Abstract

A new moving object contour detection approach is proposed in this paper. This approach combines still image segmentation and motion segmentation techniques together to provide fast processing speed and accuracy of object contour. The motion segmentation algorithm used in the proposed approach is the hierarchical adaptive structured mesh (HASM), which extracts the motion boundary of the moving objects and removes noise regions. Then inside the motion boundary, the foreground pixels are detected and the morphological operations are applied to produce foreground blobs. Simple edge detection technique is used to get the continue contour of the moving object. The combination of these two techniques reduces the whole computational cost of traditional contour extraction approaches and efficiently extracts the moving object contour