基于C-均值聚类的视觉监控背景图像构建算法

A Background Reconstruction Algorithm Based on C-means Clustering for Video Surveillance

运动分割是视觉监控系统的关键环节,背景图像减除法是实现运动分割的常用方法,但需要静态背景的参考图像来初始化背景模型。而建立背景参考图像时,常难以控制进入监视区域的人或物体,不能净空场景,因此必须从动态场景中构建一帧反映被监视场景固定构成的背景参考图像。论文提出了一种基于C-均值聚类的视觉监控背景图像构建算法,实时构建灰度或彩色背景图像,实验验证了该算法的有效性。

Motion segmentation is the key step for video surveillance,and background subtraction method is mainly used for this purpose.To do so,a reference background image is needed to initialize the background models.But actually it is difficult or impossible to control the area being monitored.This paper presents a static background image reconstruction algorithm based on C-means clustering for video surveillance.The algorithm runs real-time and does not require a large memory space in which to store the image sequences.Our experimental results show the feasibility and effectiveness of the proposed algorithm.