一种视频序列的超分辨率重建算法  

A Superresolution Reconstruction Algorithm of Video Sequence

作　　者：韩玉兵[1] 陈小蔷[1] 吴乐南[1] 

机构地区：[1]东南大学无线电工程系,江苏南京210096

出　　处：《电子学报》2005年第1期126-130,共5页Acta Electronica Sinica

摘　　要：本文构建了一种视频序列超分辨率重建框架 .在此框架下 ,讨论了基于最小二乘规整化泛函的单帧图像的超分辨率重建算法及其收敛性、凸性和参数选择等 ;还提出了基于加权矩阵的运动配准融合 ,并研究了运动补偿阵和加权阵的构成和特点 .仿真结果表明方法的有效性和实用性 .A framework of reconstructing a superresolution video sequence from a low resolution video sequence is proposed in this paper.Under this framework,we discuss a superresolution restoration algorithm from a single frame based on regularized least mean square functional.Its iteration formula,convergence,convexity and the choice of parameters are thoroughly studied.We also present a registration algorithm based on weight matrices.The characteristics of motion compensation matrices and weight matrices are discussed in detail.Experimental results demonstrate the power of the proposed method.

关 键 词：视频处理 超分辨率重建 Tikhonov规整 运动补偿 图象配准 

分 类 号：TN911.73[电子电信—通信与信息系统]