基于均值场理论和马尔可夫场的运动目标分割算法

研究了基于均值场理论和马尔可夫场的运动目标分割方法.该算法先对帧差图像进行简单的前背景划分,再采用均值场理论(MFT),建立马尔可夫随机场(MRF)模型,构造系统相应的能量函数.然后通过求取最大后验估计(MRF-MAP),即求最小能量函数,得出标记场,提取运动目标.实验结果证明:该算法能够很好地消除噪声,对运动目标分割具有较好的分割效果.

Segmentation Algorithm of Moving Object Based on Mean Field Theory and MRF-MAP

In the paper,moving object segmentation algorithm based on mean field theory (MFT) using maximum a posteriori estimation of Markov random field (MRF-MAP) is proposed.This algorithm begins with a simply foreground image derived from the difference between two frames,Computing the energy function of Markov random field (MRF) is constructed by applying the MFT.The MRF-MAP which is minimum energy function,is adopted to obtain the label field and extract the moving object.Experimental results are provided using MPEG-4 test sequences.The results show that the new method detects changes accurately and is robust to noise.