摘要
自适应字典学习利用图像结构自相似性,将图像自身作为训练样本,通过字典学习使图像中的相似块在字典下具有稀疏表示形式.本文将全局字典学习中利用图像库获取附加信息的思想融入到自适应字典学习的过程中,提出了一种基于自适应多字典学习的单幅图像超分辨率算法,从低分辨率图像自身与图像库同时获取附加信息.该算法对低分辨率图像金字塔结构中的图像块进行聚类,在聚类结果的引导下将图像库中的图像块进行分类,利用各类中的样本分别构建针对各类的多个字典,从而确定表达重建图像块的最优字典.实验表明,与Sc SR、SISR、NLIBP、CSSS以及m SSIM等算法相比,本文算法具有更好的超分重建效果.
Adaptive dictionary learning uses the lowresolution image itself as training samples to make the similar patches have sparse representation over the learned dictionary,so that extra information can be exploited from structural selfsimilarity by dictionary learning. In this paper,we propose a single image super resolution method based on adaptive multidictionary learning. To exploit extra information from both the lowresolution image itself,and the image database,the proposed method incorporates the idea of global dictionary learning that the image database can be used to obtain extra information into the process of adaptive dictionary learning. In the proposed method,all patches in the image pyramid of the lowresolution image are clustered into several groups,then each patch satisfying a certain condition in the database is classified into one of these groups with the supervision of the clustering results,and multi-dictionary learning is used to learn corresponding dictionaries for different groups. Experimental results demonstrate that our method achieves better result compared with ScSR,SISR,NLIBP,CSSS and m SSIMmethods.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2015年第2期209-216,共8页
Acta Electronica Sinica
基金
国家科技支撑计划(No.2012BAH31B01)
国家自然科学基金(No.61171117)
北京市教育委员会科技计划重点项目(No.KZ201310028035)
关键词
超分辨率
稀疏表示
自适应字典学习
全局字典学习
super resolution
sparse representation
adaptive dictionary learning
global dictionary learning
