基于稀疏表示的WMSN红外和可见光图像融合  

Infrared and visible image fusion in WMSN based on sparse representation

作　　者：罗晖[1] 刘洁丽[1] 祁美丽[1] 

机构地区：[1]华东交通大学信息工程学院,江西南昌330013

出　　处：《计算机工程与科学》2014年第5期958-962,共5页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(61261040)

摘　　要：在使用无线多媒体传感网络WMSN进行环境监测的过程中,对同一场景所采集的红外和可见光源图像进行信息融合时,传统的方法融合的数据量较大且没有充分考虑其内在稀疏性和丰富的结构特征,图像融合的质量不高。将稀疏表示理论应用于WMSN红外和可见光图像融合中,在原始DCT冗余字典基础上,结合K-SVD字典训练算法和同步正交匹配追踪SOMP算法对WMSN红外和可见光图像进行有效的稀疏表示,并选择自适应加权平均融合规则对稀疏表示系数进行融合处理。仿真结果表明,相对于传统的基于空域及变换域的红外和可见光图像融合方法,该方法更能从WMSN含噪图像中有效地保留源图像的有用信息,获得较好的融合效果。When Wireless Multimedia Sensor Network （WMSN） is used for environment detection and the infrared image and the visible image collected from the same scene are fused, the traditional ap- proaches have large amount of fused data and do not fully consider the internal sparsity and the complex- ity of structure features, so the fusion quality is low. The theory of sparse representation is applied to WMSN infrared and visible image fusion. Based on the original DCT redundant dictionaries, the K-SVD training method is combined with the Simultaneous Orthogonal Matching Pursuit （SOMP） algorithm to do effective sparse representation for WMSN infrared images and visible images. And adaptive weighted average fusion rule is selected to deal with the sparse representation coefficients. Experimental results show that, compared with traditional infrared and visible image fusion methods based on spatial and transformed domains, the proposed method can effectively preserve the useful information and get the better fused image.

关 键 词：无线多媒体传感器网络 图像融合 稀疏表示 同步正交匹配追踪 自适应加权平均 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]