一种改进的多帧图像超分辨率重建算法  

An Improved Superresolution Reconstruction Algorithm Based on Multiframe Image

作　　者：刘峰[1] 朱秀昌[2] 

机构地区：[1]南京邮电大学图像处理与图像通信江苏省重点实验室,江苏南京210003 [2]南京邮电大学通信与信息工程学院,江苏南京210003

出　　处：《南京邮电大学学报（自然科学版）》2007年第4期19-23,共5页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：江苏省自然科学基金(KB2004151)资助项目

摘　　要：文中构建了超分辨率重建图像的一般框架。在对图像模糊的不确定性和复杂性作一定限制条件下,讨论采用最小二乘方规整化方法重建除运动外其它因素引起降质的低分辨率图像;并进一步提出了采用改进的递归最速下降迭代算法实现多帧图像的超分辨率重建。计算机模拟结果表明,该方法具有较好的重建图像质量。A general framework of reconstructing a super-resolution image is proposed in this paper. Under the some restriction condition of degradation image uncertainty and complexity, we discuss the least mean square regulation reconstruction method of low resolution images which result from various factors except motion, and farther present an improved recursive steepest-descend iteration algorithm to realize the super-resolution reconstruction of more than one low resolution frame. Experimental results illustrate that this method can improve image reconstructing quality.

关 键 词：图像插值 超分辨率重建 最小二乘方规整 

分 类 号：TN919.8[电子电信—通信与信息系统]