基于块估计的运动目标检测方法  

Detecting method for moving targets based on block estimation

作　　者：陈忠碧[1] 张启衡[1] 彭先蓉[1] 任臣[1] 

机构地区：[1]中国科学院光电技术研究所

出　　处：《光电工程》2006年第6期15-19,共5页Opto-Electronic Engineering

基　　金：国家863计划预研基金课题

摘　　要：提出了一种适合于运动目标检测的块运动分析方法,用以补偿移动背景所带来的杂波信息,准确检测运动目标。在对原始图像进行滤波的基础上,选择合适的子块,通过块匹配,得到背景位移矢量;利用这些运动参数,在相邻帧之间进行运动补偿,达到分割运动目标的目的。该算法对背景的适应性能好。Based on the estimation of moving block, a method for the detection of moving targets was proposed in the paper. By matching moving block to compensate the confusion brought by the moving background, moving target can be segmented correctly. By filtering, choosing suitable sub-block and matching to the raw image, the displacement vectors of the background can be obtained. Using these parameters, it is possible to segment moving target by moving compensation between the adjacent frames. Its adaptability to the background is very good.

关 键 词：运动目标 块估计 块匹配 复杂背景 目标检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]