基于自适应双边全变差的图像超分辨率重建  

A Self-Adapting Bilateral Total Variation Technology for Image Super-Resolution Reconstruction

作　　者：杨欣[1] 周大可[1] 费树岷[2] 

机构地区：[1]南京航空航天大学自动化学院,南京210016 [2]东南大学自动化学院,南京210096

出　　处：《计算机研究与发展》2012年第12期2696-2701,共6页Journal of Computer Research and Development

基　　金：国家自然科学基金项目(60905009);高等学校博士学科点专项科研基金项目(20093218120015);中国科学院遥感应用研究所;北京师范大学遥感科学国家重点实验室开放基金项目(2009KFJJ012);南京航空航天大学基本科研业务费专项科研基金项目(NS2010081)

摘　　要：超分辨率图像重建技术就是通过融合多幅变形、模糊、有噪、频谱混叠的低分辨率降质图像(或视频序列)来重建一幅高质量高分辨率图像.MAP估计算法是一种广泛使用的统计重建方法.针对标准的MAP算法引入了自适应概念,引入了图像自适应加权系数矩阵;据此给出一种基于自适应双边全变差的图像超分辨率重建算法,该方法不仅能在图像超分辨率重建过程中抑制噪声,而且能锐化图像中的边缘信息;建立了自适应重建模型并用梯度下降法推导出迭代计算公式.实验表明,该算法在收敛性和精确性上都达到了较好的效果.Super-resolution image reconstruction has recently drawn considerable attention within the research area. For some special-purpose imaging devices such as medical imaging, remote sensor imaging and video capturing, the acquired images cannot often achieve a higher resolution because of the limitation of imaging mechanism and imaging sensor. Super-resolution image reconstruction methods attempt to create a single high resolution and high quality image from multiple low resolution observations （or a video sequence） degraded by warping, blurring, noise and aliasing. So far, existing super-resolution methods are all confronted with the problems of slow convergence and expensive computation. To satisfy the requirement of real occasion applications, an effective super-resolution reconstruction algorithm is built upon the MAP framework. In the proposed algorithm, the conception of self-adapting weight coefficients matrix （SWCM） in super-resolution technology is proposed. And a new method on super-resolution based on self-adapting bilateral total variation is given. The method takes into account respective characteristic about each LR image. It can not only sharpen edges but also help to suppress noise in the estimated HR image. The super-resolution reconstruction model and iterative scheme are developed to get the more accurate image. Experimental results using both real and synthetic data show the effectiveness of the proposed algorithm.

关 键 词：图像重建 超分辨率 自适应 双边全变差 分辨率增强 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]