多视频的时空超分辨率重建算法  

A Space-Time Super-Resolution Reconstruction Algorithm of Video Sequences

作　　者：宋海英[1,2] 何小海[1] 陈为龙[1] 孙琰玥[1] 

机构地区：[1]四川大学电子信息学院,成都610064 [2]成都电子机械高等专科学校通信工程系,成都610031

出　　处：《北京邮电大学学报》2011年第4期85-88,共4页Journal of Beijing University of Posts and Telecommunications

基　　金：教育部科学技术研究重点资助项目(107094);国家自然科学基金项目(61071161)

摘　　要：Shechtman E等提出的时空超分辨率重建算法,在输入低分辨率视频序列足够多的情况下,重建效果非常好.但在实际应用中很难获得针对同一动态场景的大量低分辨率视频序列,导致该重建算法的应用有了一定的局限性.针对这一问题,提出了一种新的时空超分辨率重建算法.采用Shechtman E等提出的算法重建出一个高时间分辨率视频序列,采用改进的迭代反投影算法提高该高时间分辨率视频序列的空间分辨率,从而得到一个高时空分辨率视频序列.实验结果表明,提出的新算法能较好地实现用较少的低分辨率视频序列重建一个高时空分辨率视频序列.With enough input of low resolution video sequences, the algorithm of space-time super-resolution reconstruction proposed by Shechtman E et al. can achieve satisfactory result. Because it＇s hard to obtain enough low resolution video sequences of the same dynamic scene in practice, the application of this algorithm is limited. A new algorithm of space-time super-resolution reconstruction is proposed, in which, a high temporal resolution video sequence is reconstructed by utilizing the algorithm proposed by Shechtman E et al, and the spatial resolution of the high temporal resolution video sequence is improved by utilizing the improved iterative back projection algorithm. Experiments show that the proposed algorithm can effectively reconstruct a high space-time resolution video sequence with limited low resolution video sequences.

关 键 词：多视频 超分辨率 时间分辨率 空间分辨率 重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]