Motion segmentation using Markov random field model for accurate moving object segmentation

In this paper, we propose a new motion segmentation algorithm for accurate moving object segmentation. The proposed algorithm is elaborately designed based on Markov random field (MRF) model. When motion decision is performed on each pixel in images by detection theory, false decisions such as miss and false alarm are occurred. These false decisions cause segmentation errors of moving objects. We remove these errors by using MRF model based on Bayes rule. By means of experiments on test sequences, we demonstrate that our algorithm achieves good performance with respect to the segmentation accuracy.