基于联合非负字典学习的遥感图像超分辨重建  

Super-resolution Reconstruction of Remote Sensing Images Based on Joint Nonnegative Dictionary Learning

作　　者：魏巍[1] 吴孔平[1] 郭来功[1] 秦蒙[1] 

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001

出　　处：《计算机工程》2016年第8期271-276,共6页Computer Engineering

基　　金：安徽理工大学青年教师科学研究基金资助项目(QN201311);安徽理工大学大学生创新创业训练计划基金资助项目(AH201410361178)

摘　　要：针对图像的超分辨重建问题,提出一种基于联合非负字典学习的单幅图像超分辨重建算法,并将其用于遥感图像的超分辨重建。利用已有的高分辨图像,通过预处理得到高低分辨样本集。给出联合非负字典学习技术,并采用该技术对高低分辨样本集训练得到稀疏的高低分辨字典,使用高低分辨字典重建高分辨图像,并分析算法的计算复杂度。实验结果表明,与双三次插值法、联合字典训练算法、耦合字典训练算法相比,该算法在保证较好重建效果的同时需要较小的计算量。Aiming at the super-resolution reconstruction of images, a super-resolution reconstruction algorithm for a single image based on joint nonnegative dictionary learning is proposed in this paper, and it is applied in the super- resolution reconstruction of remote sensing images. Using existing high-resolution images, high-resolution and low- resolution samples are obtained by preprocessing. Joint nonnegative dictionary training technology is proposed,and high- resolution dictionary and low-resolution dictionary are obtained by training high-resolution and low-resolution samples, respectively. Super-resolution remote sensing image is recovered by these dictionaries, and computation complexity is analyzed. Experimental results show that, compared with bicubic interpolation method, Joint Dictionary Training （JDT） algorithm and coupled dictionary training algorithm, the proposed algorithm requires lower computation cost to achieve better reconstruction effect.

关 键 词：联合字典训练 稀疏表示 超分辨重建 遥感图像 计算复杂度 

分 类 号：P391.41[天文地球—地球物理学]