图像序列的快速超分辨率恢复算法  

A Fast Super Resolution Method from Image Sequences

作　　者：田岩[1] 王志成[2] 柳健[1] 

机构地区：[1]华中科技大学电子与信息工程系,湖北武汉430074 [2]图像信息处理与智能控制国家教育部重点实验室,湖北武汉430074

出　　处：《电子学报》2004年第12期2074-2077,共4页Acta Electronica Sinica

基　　金：国家自然科学基金 (No .60 2 72 0 99);国家 863计划项目 (No.2 0 0 2AA40 1 0 0 1 4B) ;测绘遥感信息工程国家重点实验室开放基金 (No .WKL(0 2 ) 0 1 0 3)

摘　　要：图像序列的超分辨率恢复是目前国际图像复原界的一个研究热点 .本文提出了一种基于分层机制的图像序列超分辨率恢复方法 .在将问题采用正则化方法变为适定问题后 ,提出了一种级联模板的快速运算方法 .The super-resolution technique from image sequence is an active research issue in the field of image restoration.For this problem, a novel method based on delaminating strategy is described in this paper.The delaminating strategy divides the original problem into several sub-problems.Each of these sub-problems is only connected to one degrading factor.In addition, it is suggested that the ally-template algorithm,which is implemented by a serious mask operator,can resolve each sub-problem.To show the effectiveness of the presented algorithm,the simulative results carried out on an aerial image sequence are compared with those of conventional methods such as the maximum a posteriori probability (MAP) method and the projections onto convex sets (POCS) method.Furthermore,the complexity of the new method is discussed.The experimental result prove this method to be efficient.

关 键 词：图像序列 超分辨率 恢复算法 级联模板 图像复原 快速 恢复方法 正则化方法 分层 运算方法 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]