系统工程与电子技术
• 软件、算法与仿真 • 上一篇 下一篇
石春雷1,2, 杨文佳2, 黄培康3
出版日期:
发布日期:
SHI Chun-lei1,2, YANG Wen-jia2, HUANG Pei-kang3
Online:
Published:
摘要:
针对运动相机中视频的随机抖动和低频晃动问题,提出了一种基于范数优化路径的视频稳像算法。该算法在特征点粗匹配之后引入概率进化抽样一致性算法剔除误匹配点,避免了局部运动对全局运动估计的影响,然后在定制的L1范数约束下构建相机的最优化运动路径,最后利用新的路径进行运动矫正获取稳定的视频图像序列,实验结果验证了算法的有效性。
Abstract:
To deal with the problem of random shakiness and low frequency bounces, an improved video stabilization method based on norm optimal camera paths is proposed. Probability evolution sample consensus algorithm is used to eliminate the error matching points after coarse matching of feature points, avoiding the interference of local motion. Then camera paths are gained by L1 norm optimization. Finally, image warp is computed to get stabilized video sequences. Experimental results for different video sequences show the proposed method has a better veracity of stabilized results.
石春雷, 杨文佳, 黄培康. 基于范数优化路径的视频稳像算法[J]. 系统工程与电子技术, doi: 10.3969/j.issn.1001-506X.2015.06.28.
SHI Chun-lei, YANG Wen-jia, HUANG Pei-kang. Video stabilization algorithm based on norm optimal paths[J]. Systems Engineering and Electronics, doi: 10.3969/j.issn.1001-506X.2015.06.28.
0 / / 推荐
导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks
链接本文: https://www.sys-ele.com/CN/10.3969/j.issn.1001-506X.2015.06.28
https://www.sys-ele.com/CN/Y2015/V37/I6/1411