基于POCS的超分辨率重建算法研究  

Research of reconstruction algorithm based on POCS super-resolution

作　　者：杨大伟[1] 李丹[1] 李健[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《微型机与应用》2014年第8期39-41,44,共4页Microcomputer & Its Applications

基　　金：黑龙江省科技攻关项目(GC12A305);中央高校自由探索计划(HEUCF130801)

摘　　要：针对传统的凸集投影(POCS)算法重建后的结果图像存在边缘模糊的问题,提出了一种通过小波变换与分形插值得到高分辨率初始图像的估计方法。该方法通过对一幅图像进行小波分解得到低频重构图像和高频重构图像,对高频重构图像使用分形插值保留了图像的纹理和边缘信息。仿真实验结果表明,该方法可行有效,改善了图像的边缘特性及整体质量,与传统的POCS算法相比,本文方法提高了重建图像的峰值信噪比。For the weakness of conventional POCS algorithm, the image edge of POCS is blurry, an improved initial estinmtion algorithm based on wavelet interpolation is proposed. In which the wavelet and fraetal interpolation is combined to get the initial image, and we get the high frequency image and the low frequency by wavelet, then keep the texture and edge information using fractal interpolation. Experiments show that this proposed algorithm improves the edge characteristic of the super-resolution image. Furthermore, the new algorithm improves the PSN1R of the the super-resolution image .

关 键 词：超分辨率 凸集投影 小波变换 分形插值 

分 类 号：TN911.73[电子电信—通信与信息系统]