边缘增强的多字典学习图像超分辨率重建算法  

Image Super-Resolution Based on Edge-enhancement and Multi-dictionary Learning

作　　者：詹曙[1,2] 方琪[1] 

机构地区：[1]合肥工业大学计算机与信息学院,合肥230009 [2]安全关键工业测控技术教育部工程研究中心,合肥230009

出　　处：《光电工程》2016年第4期40-47,共8页Opto-Electronic Engineering

基　　金：国家自然科学基金资助(61371156);安徽省科技攻关计划(1401B042019)

摘　　要：针对目前基于字典学习的图像超分辨率算法中边缘保持能力有限、易产生视觉伪影等不足,本文提出了一种边缘增强的多字典学习图像超分辨率重建算法,可以有效的恢复图像边缘细节信息。算法首先对训练图像块进行聚类处理,然后使用Boost K-SVD算法快速学习多组字典对,超分辨重建时自适应选择最优字典对进行稀疏分解和重建。为了改善重建后图像的边缘视觉效果,根据输入的低分辨率图像引入边缘方向保持正则项,同时学习自然图像库的边缘锐度统计先验对重建后图像的边缘进行约束。实验结果验证了本文算法的有效性。In order to overcome the weak of the limit ability of preservation of edges and easy to produce visual artifacts in some super-resolution methods based on dictionary learning,we propose multi-dictionary learning image super-resolution method with edge-enhanced,which can effectively restore the image edge details.Firstly,the training image patches will be classified by using K-means,and then quickly learn multi-dictionary pairs by employing the Boost K-SVD algorithm.During the super-resolution reconstruction,the method adaptively selects the optimal dictionary pairs for sparse decomposition and recovery.To improve the visual quality of edge after image reconstruction,we employed direction-preserving regularization according to the input test low-resolution（LR） image,meanwhile learning the natural image database edge sharpness statistics prior to constraint the image reconstruction of edges.The experimental results demonstrate the effectiveness of the proposed algorithm.

关 键 词：超分辨率重建 字典学习 统计先验 边缘增强 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]