基于兴趣区域的特征匹配的电子稳像方法  

Electronic Image Stabilization Based on Feature Matching of ROI

作　　者：吉淑娇[1] 雷艳敏[1] 朱明[2] JI Shujiao LEI Yanmin ZHU Ming(School of Electronic and Information Engineering, Changehun University, Changehun 130022, China Changehun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China)

机构地区：[1]长春大学电子信息工程学院,长春130022 [2]中国科学院大学长春光学精密机械与物理研究所,长春130033

出　　处：《吉林大学学报（信息科学版）》2016年第3期390-395,共6页Journal of Jilin University（Information Science Edition）

基　　金：吉林省教育厅基金资助项目(2015LY502L11);吉林省科技发展计划基金资助项目(201401084JC)

摘　　要：为提高电子稳像技术中运动矢量估计的准确性,采用基于兴趣区的特征点匹配进行运动矢量估计。该算法选取图像的兴趣区(ROI:Region of Interest)进行Harris的特征点检测,将ROI内的特征点稀疏化,利用SAD准则在相邻帧的特征窗内寻找匹配点对和距离准则剔除误匹配点进行逐帧运动补偿。实验结果验证了该算法的有效性。The key technology of electronic image stabilization is the motion vector estimation. FPS （Featurepoints） matching based on the ROI （ Region of Interest） area is used to estimate motion vectors. Firstly, R0I ofthe input image was selected, then the FPS were detected by Harris. Secondly, the rule of SAD inside thewindow was used to sparse the FPS and find matching point in the adjacent frame. Finally, distance criterion wasapplied to eliminate false match points. Results of frame by frame motion compensation show the effectiveness ofthe algorithm.

关 键 词：兴趣区域 特征窗 距离准则 

分 类 号：TP139.4[自动化与计算机技术—控制理论与控制工程]