基于像素筛选技术的光流估计方法  

Pixel-filtrated based method for determining optical flow

作　　者：马龙[1] 王鲁平[1] 李飚[1] 陈小天[1] 

机构地区：[1]国防科技大学ATR重点实验室,长沙410073

出　　处：《吉林大学学报（工学版）》2012年第4期979-984,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：国防预研基金项目

摘　　要：针对传统的光流估计方法对光照变化敏感,且在物体边缘处难以获得准确的光流估计的问题,提出了一种基于像素筛选技术的光流估计方法。该方法在光变环境下定义了基本方程,并采用了改进的局部光流技术。新的局部光流技术首先对局部窗口内的像素进行两次筛选,然后基于保留像素采用加权最小二乘法估计光流。光变、噪声环境下的测试实验证明了该方法的有效性和鲁棒性。The traditional methods for determining optical flow are not robust under variable illumination, and do not work well at object edges. To overcome these shortcomings, a pixel-filtration based method is proposed. The method defines a new basic equation under variable illumination, and uses an improved local flow technique. First, this technique selects proper pixels twice from the local window then, it determines the flow using the weighted-least-square method based on the selected pixels. Experiment results under variable illumination and noise demonstrate the validity and robust of the proposed method.

关 键 词：计算机应用 光流 边缘 光照变化 噪声 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]