HOS运动目标分割算法在视频监控中的应用

为了提高视频监控中运动目标分割的速度和准确度,研究并实现了一种基于高阶统计量HOS(HigherOrder Statistics)的分割算法.首先根据HOS假设检验处理帧差图,判定像素点是否属于运动区域,阈值通过灰度共生矩阵获得,考虑了背景纹理的慢变化.然后,用矩形框聚类法大致确定运动目标的范围,在该范围内使用形态运算法和首尾扫描法去除空洞.最后,使用模板相与法获得帧图像的运动目标模板,从原图像中分割运动区域.算法采用了由粗到精的分析策略,实验表明,是一种快速稳健的算法.

Application of HOS based Moving Object Segmentation Algorithm in Video Surveillance

In order to improve the efficiency of moving object segmentation in the video surveillance, a HOS (Higher Order Statistics) based algorithm is proposed and implemented. Firstly, a higher order statistics hypothesis testing in inter-frame difference is used to automatically determine moving pixels in a general video sequence. Additionally, the threshold is found by gray level co-occurrence matrix considering the background texture change. Secondly, motion regions are extracted using a matrix-cluster approach, and then, morphologic and head-tail scan methods are used to fill the regions. Thirdly, for three consecutive frames, the moving object template in the second frame can be found according to the templates in the two inter-frame difference images. Based on this template, motion regions in the second frame can be segmented., improves the process speed and results effect. Experimental results demonstrate that the algorithm in this paper which is using coarse-to-fine technique is effective to segment moving object in video sequence.