基于改进SIFT算法的视频图像序列自动拼接  

Automatic video mosaic imaging based on improved SIFT algorithm

作　　者：卢斌[1] 宋夫华[1] 

机构地区：[1]中国计量学院机电工程学院,杭州310018

出　　处：《测绘科学》2013年第1期23-25,共3页Science of Surveying and Mapping

摘　　要：本文提出了一种自适应帧采样和限定特征提取区域的拼接方法,根据帧间重叠率和帧间隔建立线性模型,并把各帧图像对准到其前后的关键帧上。在特征点提取方面,提出了一种改进的SIFT算法进行特征点提取,并采用随机采用一致性(RANSAC)方法来更新匹配点,在图像融合中采用线性加权渐入渐出的自然融合算法。实验结果表明:该方法对一般场景能稳定的抽取到关键帧,并进行拼接,取得了较好的拼接效果。As the existing methods for video image mosaic take high computational costs and huge computation, this paper proposed a method for making panoramas by getting an adaptive frame and matching method of limiting behavior region. According to interframe rate and frame overlap interval of linear model, all frames were matched to their latest neighbor key frames. In the proeess of mosaic, at the aspect of feature extracting, this paper used the improved SIFT feature operator, and the improved RANSAC algorithm to reduce image registration error. This article used the linear weighting fusion algorithm to achieve the desired gradually-enter gradually- leave effeet in the overlap image region. Experiments on video sequenees showed that this algorithm could extract mosaic key frames with satisfying qualily.

关 键 词：视频图像 重叠率 SIFT 图像融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP751[自动化与计算机技术—计算机科学与技术]