复杂背景下运动目标分割算法研究

视频监控中,准确而快速地分割出运动物体是提取各种信息的前提和关键.以图像差分法为基础,提出了一种在复杂背景下有效分割运动目标的方法.运用改进的Surendra算法提取背景.对连续帧图像的R,G,B三通道帧差分图像采用PCA融合和二值形态学重构,以提取和更新背景.将粗分割图像转换到HSV域中,采用V分量阈值法消除阴影,并应用彩色投影法解决连通体粘连和路面反光问题.实验结果表明该方法能够结合各种算法的优势,快速而较准确地提取出运动车辆目标.

Segmentation of Moving Targets in Complex Background

Segmenting the targets correctly and quickly in video surveillance is the precondition for extracting all kinds of information.An algorithm of segmenting moving targets in complex background based on difference methods is proposed in this paper.The background is extracted by modified Surendra algorithm.The differences of two consecutive frames in channel R,G and B,converted into binary images,are fused by Principal Component Analysis method and morphological processing is applied to them to extract and upda...