基于自训练字典学习的单幅图像的超分辨率重建  

Single Image Super-resolution Reconstruction Based on Self-learning Dictionary

作　　者：张强[1] 张爱梅[1] 王华敏[1] 陈鹏[1] 

机构地区：[1]郑州大学机械工程学院,河南郑州450001

出　　处：《红外技术》2015年第9期736-739,共4页Infrared Technology

摘　　要：针对单幅低分辨率图像的超分辨率重建问题,提出了一种基于自训练字典学习的超分辨率重建算法。首先根据图像的退化模型,对输入的低分辨率图像进行降质处理,然后利用K-SVD方法训练字典,获得重建所需要的先验知识,最后根据先验知识重建高分辨率图像。仿真实验的结果表明,利用该方法获得的高分辨率图像在视觉效果和客观评价上均优于传统方法,同时算法的时间效率也有很大的提升。Based on the self-learning dictionary, a super-resolution reconstruction method of single image is proposed. First of all, according to the image degradation model, the low-resolution image input is processed with blurred and downsampled operations. Then the dictionary is trained with K-SVD method, and we obtain the priori knowledge for reconstruction. Finally, the high-resolution image is reconstructed based on the priori knowledge. The result of simulation experiment shows that the method is superior to conventional methods in the visual effects and objective evaluation, and the time efficiency of the algorithm is also significantly improved.

关 键 词：超分辨率重建 稀疏表示 自训练字典学习 K-SVD 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]