基于R滤波器的超分辨图像配准算法  

A Registration Algorithm for Super-resolution Images Based on R Filter

作　　者：李凤霞[1] 张玲玉[1] 周杨[1] 

机构地区：[1]佳木斯大学信息电子技术学院,黑龙江佳木斯154007

出　　处：《内蒙古大学学报（自然科学版）》2010年第4期443-449,共7页Journal of Inner Mongolia University：Natural Science Edition

基　　金：黑龙江省教育厅科学技术项目(编号11531388)资助

摘　　要：图像配准是超分辨率图像重建的关键.提出了一种基于R滤波器的超分辨率图像重建的三步算法.第一步:对原图进行加窗模糊;第二步:根据全局运动模型配准,计算平移参数和旋转角度,对图像进行平移和旋转;第三步:把序列低分辨率图像配准到同一个坐标系中,重构高分辨率图像.实验表明,三步算法的配准精度和超分辨率图像重建效果较好.Image registration is the key to super-resolution image reconstruction.A three-step registration algorithm for super-resolution images based on R filter is presented.The first step is to obfuscate the original image by adding windows.The second step is to revolve and translate the image by calculating translation parameters and rotation angle according to the requirement of the global motion model.The last step is to match image sequences of low resolution in the same coordinate system and to reconstruct the high resolution images.Experiment results show that matching accuracy and result of reconstructing super-resolution images are greatly improved.

关 键 词：图像配准 图像重建 超分辨率 三步法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]