基于分段处理的视频稳像算法  

Algorithm of video stabilization based on segment processing

作　　者：任钰伟 伍新华[1] 刘钢[1] 

机构地区：[1]武汉理工大学计算机科学与技术学院,武汉430063

出　　处：《计算机应用研究》2017年第10期3174-3176,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(51179146)

摘　　要：针对视频拍摄过程中出现抖动的问题,根据在短时间内摄像机运动轨迹保持不变的观察现象,提出了一种按时间分段处理视频的视频稳像(SPVS)方法。首先提取视频的SURF特征点,用u MLESAC算法快速准确地去除异常点,并使用最小二乘法求解帧间仿射运动模型;然后利用分段方法优化处理视频的仿射运动模型参数;最后利用视频补偿得到稳定的视频文件。实验表明SPVS方法可有效地去除视频中的抖动成分。Aiming at the problem that there is jitter existing in the process of video shooting, this paper proposed an algorithm of video stabilization based on segmentation processing （SPVS） , due to the observation phenomenon that the camera motion track stayed the same in a fixed time. At first,it extracted features by SURF（ speed up robust feature） method, used the uM- LESAC（ an adaptive random sample consensus method） to exclude the outlier, and estimated the affiue transform parameters through the least square method. Next, it used the SPVS method to smooth the original video motion. At last,it achieved a stable video sequence using video compensation. Experiment results show that this method is able to effectively remove the jitter corn- ponents in the video.

关 键 词：视频抖动 视频稳像 仿射模型 视频分段 运动平滑 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]