一种基于KLT-RANSAC全局运动估计的电子稳像算法研究  

Research on Electronic Image Stabilization Based on KLT-RANSAC Global Motion Estimation

作　　者：王智慧[1] 王敬东[1] 李鹏[1] 张春[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016

出　　处：《光电子技术》2012年第1期46-51,55,共7页Optoelectronic Technology

基　　金：国家自然科学基金资助项目(61074161);江苏省重点实验室开放基金资助(YCCK200901)

摘　　要：全局运动估计是电子稳像的核心,很大程度上影响整个稳像系统的性能。提出了一种基于KLT-RANSAC全局运动估计的电子稳像算法,该方法首先通过增加距离约束机制,并进行亚像素级定位,优化Harris特征点,选择KLT快速匹配算法实现特征点的粗匹配,之后引入了RANSAC算法剔除误匹配点,去除视频序列中分布在运动物体上的特征点,避免了局部运动对全局运动估计的影响。实验结果表明:本算法可以有效解决特征点匹配过程中的误匹配问题,避免局部运动对稳像的干扰,快速准确地计算出全局运动参数,最终可以实时处理存在平移和旋转抖动的视频,输出稳定高质量的视频。Global motion estimation is the kernel of electronic image stabilization,greatly influencing the system performance.An electronic image stabilization algorithm is presented based on KLT-RANSAC global motion estimation.The feature points distributing relative uniform are obtained by optimizing Harris corner detection algorithm.The positional precision is improved to the sub-pixel level.Coarse matching of feature points is realized by KLT fast matching algorithm, and then RANSAC algorithm is used to eliminate the error matched points which distribute on the moving objects,avoiding the interference of local movement on the image stabilization.Experimental results show that the algorithm can effectively solve the mismatch problems and finally avoid the interference of local movement,calculate the global motion parameters quickly and accurately,and,process stable and high-quality video containing translation and rotation shake in real-time.

关 键 词：全局运动估计 随机抽样一致性 特征点匹配 误匹配 稳像 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]