超分辨率重建算法综述  

Survey of super-resolution algorithms

作　　者：李欣[1] 崔子冠[1] 朱秀昌[1] 

机构地区：[1]南京邮电大学图像处理与图像通信实验室,江苏南京210003

出　　处：《电视技术》2016年第9期1-9,共9页Video Engineering

基　　金：国家自然科学基金青年基金项目(61501260);江苏省自然科学基金项目(BK20130867;BK20140891);江苏省普通高校研究生科研创新计划项目(CXLX12_0474)

摘　　要：随着高分辨率移动设备和超高清电视的发展,对已有的低分辨率视频进行超分辨率上采样成为最近的一个研究热点。对已有的超分辨率重建算法根据输入输出方式的不同,分为多图像超分辨率重建、单图像超分辨率重建、视频超分辨率重建三大类,综述了其中每类算法的发展情况及常用算法,并对不同算法的特点分析比较。随后讨论了多图像超分辨率重建和单图像超分辨率重建方法对视频超分辨率重建方法的影响,最后展望了超分辨率重建算法的进一步发展。With the development of high-resolution mobile devices and ultra-high-definition televisions, applying super-resolution up-sampling to the existing low-resolution video has become a hot research topic recently. In this paper, according to the different way of input and output, the existing super-resolution reconstruction algorithms are divided into three categories：multiple-image super-resolution reconstruction, single-image super-resolution reconstruction, and video super-resolution reconstruction. The de- velopment and common usage of each category are reviewed, and the characteristics of different algorithms are compared. In the following, the impact of muhi-image super-resolution reconstruction and single-image super-resolution reconstruction on video su- per-resolution reconstruction is discussed. Finally, the development prospect of super-resolution reconstruction algorithm is ana- lyzed.

关 键 词：超分辨率 超高清电视 配准 稀疏表示 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.4[电子电信—信息与通信工程]