基于光流技术的运动目标检测和跟踪方法研究  

The Detection and Tracking of Moving Objects by Using Optical Flow Technique

作　　者：戴斌[1] 方宇强[1] 孙振平[1] 王亮[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,长沙410073

出　　处：《科技导报》2009年第12期55-60,共6页Science & Technology Review

摘　　要：光流技术作为一种重要的二维运动估计技术,在运动目标检测和跟踪中有着重要的作用。为了更好地将光流技术运用到实时的运动目标检测和跟踪系统中,针对微分光流信息量丰富但计算量大、特征光流计算量小但信息量不足的现状,提出了一种基于最优估计的点匹配技术和光流均匀采样策略的光流场计算方法,并通过对灰度化后的光流场进行自适应阈值分割、形态学滤波等处理,实现了实时的运动目标检测和跟踪。通过对图像序列取700个样本点的仿真实验表明,该方法帧间处理时间基本小于100ms,同时基本解决了信息量和计算量的矛盾。The Optical Flow Technique （OFT）, as an important technique for estimating a two dimensional movement, is widely used in Moving Object Detecting and Tracking （MODT）. However, the OFF method usually suffers from a heavy calculation load and inadequacy in information. In this paper, we propose an OFT method which is based on the technology of point matching and the strategy of uniform sampling and can be used in MODT in a real time through graying optical flow field. The simulation experiment with samples of 700 points per image indicates that the frame processing time is less than 100 ms, which means an adequate amount of information handled with reasonable computation cost.

关 键 词：微分光流 特征光流 运动目标检测 运动目标跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]