Segmentation, Based On The Edge Of The Motion Picture

Motion segmentation is the process of dividing video frames into regions that have different motions, providing a cut-out of the moving objects. Such segmentation is a necessary first stage in many video analysis applications, but providing an accurate, efficient motion segmentation still presents a challenge.This thesis proposes a novel approach to motion segmentation, using the edges in a frame. Edges of the image are detected out firstly, labeled according to the motion that they obey then and the areas of the frame between edges are divided into regions. At last, using the Bayesian framework presented determines the most likely region labeling and depth ordering with the labeled edges. An efficient implementation of this framework is presented, for segmenting two motions (foreground and background) using two frames. The Expectation-Maximization algorithm is used to determine the two motions and calculate the label probability for each edge. The best motion labeling for these regions is determined using simulated annealing.The results demonstrate that the proposed edge-based approach is an accurate and efficient method of obtaining a motion segmentation.