基于ASIFT的图像拼接技术在监控系统中的应用  

Image Stitching Technology Application Based on ASIFT in Monitoring System

作　　者：李校林[1,2] 刘欣[1] 谢灿 

机构地区：[1]重庆邮电大学,重庆400065 [2]重庆信科设计有限公司,重庆400065

出　　处：《光电子技术》2014年第3期150-153,157,共5页Optoelectronic Technology

基　　金：科技型中小企业技术创新基金(11C26215113601)

摘　　要：提出了一种基于改进仿射-尺度不变特征转换(ASIFT)特征的监控图像拼接算法,ASIFT特征比传统的尺度不变特征变换(SIFT)算法具有完全的仿射不变性,在低重叠图像匹配上,ASIFT算法能够检测到更多准确的特征点,满足拼接要求。实验结果表明,该算法在视频监控系统中具有很强的实用性。In order to meet a wider range of regional monitoring requirements, video surveil- lance systems generally use multiple stationary cameras to monitor the regional scene and, for different cameras capture video frames, use the video frame image stitching technology for pano- rama stitching and demarcate the interested target. In order to meet the monitoring requirements of a large area, a feature--based ASIFT monitor image stitching algorithm is proposed with com- parison with the traditional SIFT algorithm. It is shown that ASIFT have fully affine invariant. Therefore, as the image matching in the low overlap, more accurate feature points can be detec- ted with ASIFT algorithm, while stitching requirements are met. It is experimentally shown that the algorithm in video surveillance system is very practical.

关 键 词：标定 仿射-尺度不变特征转换 图像拼接 尺度不变特征变换 视频监控系统 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]