摘要
复杂背景下运动分析首先需要进行背景运动参数估计。提出一种采用KLT算子并结合Kanade-Lucas跟踪算法来计算帧间特征点的匹配,然后利用帧间特征点间的匹配关系采用RANSAC算法来进行背景运动参数估计。实验结果证明这种方法具有计算量小,能实时实现并提供可靠的背景运动模型估计。
Background affine model parameters estimation is a key problems in image registration between frames. Extended Kanade-Lucas tracker and RANSAC algorithm were combined to solve the background affine model parameters. Background features were detected using outlier voting frame. As solving an affine model only need three corresponding points, the algorithm can run in real time and generate reliable results.
出处
《计算机应用》
CSCD
北大核心
2005年第8期1946-1947,共2页
journal of Computer Applications
关键词
KLT算子
特征点
图像匹配
背景运动参数估计
Affine模型
<Keyword>kanade-lucas tracker
feature point
image registration
background motion model parameters estimation
Affine model
