摘要
为了能在超分辨率重建过程中更好地保留图像的边缘及纹理等特性,提出基于稀疏表示的正则化超分辨率重建算法.首先通过K-SVD方法得到单一的超完备字典,然后在此基础上进行改进得到高、低分辨率的字典,并在重建过程中通过自适应选取正则化参数的方法动态调节目标函数中重建误差逼近项和稀疏性约束项,从而实现超分辨率重建.通过仿真实验验证该算法能够有效地提高重建图像的质量.
In order to obtain better image edge and texture features in the process of super-resolution re- construction, a regularized super-resolution reconstruction algorithm is proposed based on sparse represen- tation. A single super-complete dictionary is obtained first by K-SVD method and then improved to get dictionaries with high and low resolutions. The terms of reconstruction error approximation and sparse constraint in objective function are dynamically adjusted by means of adaptively selecting the regularization parameter in the process of reconstruction, so that the super-resolution reconstruction is realized. It is ver- ified by simulation experiment that the reconstruction image quality can be improved with the algorithm.
出处
《兰州理工大学学报》
CAS
北大核心
2015年第2期103-106,共4页
Journal of Lanzhou University of Technology
基金
甘肃省自然科学基金(11112RJZA028)
甘肃省高校基本科研项目(1203ZTC061)
关键词
超分辨率重建
稀疏表示
K-SVD
正则化
super-resolution reconstruction
sparse representation
K-SVD
regularized