基于马尔可夫随机场的运动物体检测方法

Detection method based on Markov random field for moving object

智能监控的实现是避免和防范变电站内各种潜在危险的一种有效途径.为了更准确判定变电站工作人员的运动状态,提出一种基于高斯混合模型结合马尔可夫随机场的运动物体检测方法.在图像的HSV颜色空间通过混合高斯背景建模实现对运动物体的初步检测,采用区域性马尔可夫随机场与运动物体模板匹配实现运动物体的精确检测,并根据模板去除存在的阴影.结果表明,该方法可在变电站不同背景条件下有效检测出运动物体,为运动物体的行为分析及运动场景拼接奠定了良好的基础.

Implementation of smart surveillance is an effective way to avoid potential hazard within substation. In order to more accurately determine the state of substation staff,which was a motion object in this case,a novel detection method based on Gauss mixture model and Markov random field motion was proposed. First of all,the hue,saturation,value( HSV) color space of the images were preliminarily extracted by Gauss mixture background modeling. Then a segmentation approach based on Markov random field motion was applied to re-extract the HSV image,which was followed by shadow elimination referring to respective motion templates. Experimental results comfirmed that the proposed method was capable of precisely detecting the state of moving objects within substation in different background,which laid a sound foundation for behavior analysis of motion object and image stitching.