基于稀疏表示的自适应图像超分辨率重建算法  

Image super-resolution with adaptive regularization sparse representation

作　　者：张垚[1,2] 徐斌[3] 周尚波[1,2] 郑坚[1] 

机构地区：[1]重庆大学计算机学院,重庆400030 [2]重庆大学信息物理社会教育部重点实验室,重庆400030 [3]重庆市涪陵区职业教育中心语言教研组,重庆408100

出　　处：《计算机应用研究》2013年第3期938-941,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(611103114)

摘　　要：针对图像高分辨率重建过程中稀疏解的存在性和唯一性问题以及超分辨率图像的边缘特征和平滑噪声的关系进行了研究,提出了局部正则化参数自适应选取的方法。结合联合构造字典的算法,在重建过程中动态调整正则化参数。通过对图像的超分辨率实验证明,改进的算法具有较高的可行性,能有效平衡超分辨率图像的边缘特征和平滑噪声两者的关系,与传统的超分辨率重建算法相比,有更高的峰值信噪比。Conduct research on the existence and uniqueness of sparse resolution in super-resolution image reconstruction and the relationship of edge feature and smoothing noise of super-resolution image. In order to solve these problems, this paper pro- posed the mean of adaptive regularization. Combining with joint over-complete dictionary, it realized dynamically adjust the regularization parameter. Through super-resolution image reconstruction experiment, it demonstrates that this algorithm is working efficiently and balances edge features and smoothing noise well, which has high PSNR compared with the traditional alzorithm.

关 键 词：超分辨率 自适应正则化 联合字典 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]