基于随机抽样的快速全局运动估计

全局运动估计是稳像技术中极为重要的环节,全局运动估计算法的精度和速度将很大程度影响稳像的性能。本文提出了一种基于RANSAC鲁棒算法的全局运动估计方法,该方法通过Harris角点检测算法,找到待匹配角点,并采用相似度进行匹配,剔除外点后,用预检测的随机抽样算法得到全局运动参数。通过仿真试验表明,该方法对可见光视频、红外视频都可以适用,并能在保证精度的同时,达到实时性的要求。

A Fast Global Motion Estimation Based on RANSAC

Global motion estimation is an important segment in video stabilization.The performance of video stabilization is determined largely by the accuracy and efficiency of global motion estimation algosrithm.In this paper,a fast global motion estimation based on RANSAC is proposed.The features is obtained by Harris corner detection algorithm.Feature matching based on similarity is adopted to ensure the accuracy of this algorithm.Then global motion parameters are estimated by RANSAC algorithm with preview evaluation model.The results show that this algorithm can be applied to infrared and visual video for improving the calculation accuracy and satisfing the real-time request.