基于联合稀疏模型的高分辨率SAR与光学遥感影像融合研究  

Study on Image Fusion of High-Resolution SAR and Optical Image Based on the Joint Sparsity Model

作　　者：聂品[1,2] 汤玉奇 冯兰平[1,2] 李明[1,2] 兰翠玉 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083 [2]有色金属成矿预测与地质环境监测教育部重点实验室(中南大学),湖南长沙410083

出　　处：《现代测绘》2016年第3期1-5,共5页Modern Surveying and Mapping

基　　金：国家973计划资助项目(2012CB719903);国家自然科学基金资助项目(41401402;41301453;51479215);中国博士后科学基金资助项目(2014M552161);湖南省自然科学基金资助项目(2015JJ3150);湖南省地理国情监测项目(HNGQJC201503)

摘　　要：遥感影像融合技术能通过综合不同卫星传感器获取影像互补信息达到对地物进行提取的效果。针对合成孔径雷达(Synthetic Aperture Radar,SAR)与光学影像融合的光谱扭曲和空间细节信息丢失问题,提出一种基于联合稀疏模型的高分辨率SAR与光学影像融合方法。该方法借鉴信号联合稀疏表示思想,假定SAR与光学影像均可由共有和专有稀疏表示部分组成,分别对应影像冗余信息中的有效部分与影像互补信息:其中共有部分用离散余弦字典表示,专有部分用离散余弦字典和学习字典组成的混合字典表示。在求解出融合影像在联合稀疏模型下的稀疏系数之后重建出融合影像。实验结果表明,该方法能同时保持高分辨率SAR影像空间细节信息和光学影像光谱信息,提高了高分辨率SAR与光学遥感影像的融合效果。Remote sensing image fusion technology can integrate the complementary information of images obtained from different satellite sensors to extract spatial information.For the SAR（Synthetic Aperture Radar）and optical image fusion with the problem of spectral distortion and spatial loss,it proposed a new method of high-resolution SAR and optical image fusion based on joint sparsity model.The method was inspired by joint sparsity model,and supposed that the SAR and optical images could be both represented by a common sparse component and a specific sparse component,respectively corresponding to the effective part of redundant information and complementary information of imagery.The discrete cosine transform（DCT）dictionary was used to represent common component,while the specific sparse component was represented by the hybrid dictionary of DCT and learned dictionaries.Then the sparse coefficient of fusion image under the joint sparsity model was solved by the Orthogonal Matching Pursuit algorithm and the fused image was reconstructed.The fusion image is reconstructed by solving the sparse coefficients of the fusion image under the joint sparse model.The experiment result shows that the proposed method can effectively preserve the spectral of optical image and spatial information of SAR image simultaneously and improve the quality of fusion image.

关 键 词：稀疏表示 联合稀疏模型 影像融合 光学影像 SAR影像 

分 类 号：P237[天文地球—摄影测量与遥感]