Motion Estimation for Video Stabilization Based on Feature Points and Parameter Space Method
Motion estimation is the cruciaI step for video stabilization algorithm.There are many literatnres focus on it and various methods have been proposed.The solution of this problem is now sophisticated when only replacement is taken into account.But if the rotation between two successive frames can’t be ignored.conventional methods may lose their valid.In this paper,we aim to estimate motion parameters when replacement and rotation both exist in the video sequence.A framework based on feature points is employed.In order to find corresponding feature points in successive frames to estimate global motion parameters,a novel feature point match algorithm based on parameter space is proposed.It can endure great changes of the distribution of feature points and the processing speed is rather fast.
Parameter Space Feature Point Motion estimation video stabilization
Liu Lingqiao Fu Zhizhong Deng Zaiqiang
国际会议
The International Conference Information Computing and Automation(2007国际信息计算与自动化会议)
成都
英文
1221-1224
2007-12-19（万方平台首次上网日期，不代表论文的发表时间）