基于像素概率模型的背景分割算法

Segmenting Background Based on Probability Models of Pixel

背景分割的目的是提取出图像中感兴趣的前景区域 ,本文提出了一种基于像素概率模型的背景分割算法。该算法利用高斯混合模型描述每一被观察像素的近期色彩历史 ,根据分类原则确定当前帧中每一像素的类别 ,利用在线EM算法更新模型参数。实验结果表明 ,本文提出的算法可以鲁棒地分割出动态场景中的前景和背景。

The goal of background segmentation was to extract interesting foreground regions in images. An algorithm of segmenting background based on probability models of pixel was proposed in this paper. Firstly, adaptive pixel models were modeled to describe the recent history of color at each observed pixel. Then each pixel was classified as background or foreground according to principle of classification. Finally, parameters of models were updated using on-line EM algorithm. Experimental results showed that our approach was suitable for segmenting foreground from background in dynamic environments.