线性混合光谱模型高光谱压缩感知  

For hyperspectral compressed sensing method on linear mixed spectrum model

作　　者：王忠良 何密[2] 叶珍[3] 粘永健[2] WANG Zhongliang;HE Mi;YE Zhen;NIAN Yongjian(Department of Electric Engineering,Tongling University,Tongling 244061,China;College of Biomedical Engineering and Imaging Medicine,Army Medical University(Third Military Medical University),Chongqing 400038,China;School of Electronics and Control Engineering,Chang'an University,Xi'an 710064,China)

机构地区：[1]铜陵学院电气工程学院,铜陵244061 [2]陆军军医大学(第三军医大学)生物医学工程与影像医学系,重庆400038 [3]长安大学电子与控制工程学院,西安710064

出　　处：《遥感学报》2020年第3期277-289,共13页NATIONAL REMOTE SENSING BULLETIN

基　　金：国家自然科学基金(编号:41601344);重庆市基础科学与前沿技术一般项目(编号:cstc2016jcyjA0539);安徽高校自然科学研究重点项目(编号:KJ2019A0709);安徽省级质量工程项目(编号:2016zy126);安徽省高校优秀青年骨干人才国内外访学研修项目(编号:gxgwfx2019056);后勤科研重点项目(编号:BLJ18J005)。

摘　　要：高光谱压缩感知(HCS)对于解决机载或星载高光谱数据的存储与实时传输具有重要意义。目前,线性混合模型(LMM)已被成功应用于HCS;然而,由于光照条件、地形变化以及大气作用等的影响,所获取的地物光谱会发生扰动,从而限制了HCS重建质量的提高。在LMM基础上,通过引入光谱修正项来修正光谱扰动,提出了光谱扰动修正的LMM (SPC_LMM);在此基础上,进一步提出了基于SPC_LMM的HCS (HCS_SPC_LMM)方法。该方法在采样端仅对原始高光谱图像进行光谱维压缩采样,基于压缩采样数据,将SPC_LMM应用HCS的重建,利用交替方向乘子法(ADMM)分别估计SPC_LMM中各分量的最优值,以获得最优的高光谱图像重建质量。实验结果表明,HCS_SPC_LMM能够获得优于其他典型HCS方法的重建质量。Hyperspectral Compressed Sensing(HCS) is crucial for data storage and the real-time transmission of airborne-or spacebornebased imaging platforms. The Linear Mixing Model(LMM) has been successfully applied to HCS reconstruction. However, the obtained spectrum may be disturbed, thereby limiting the improvement of reconstruction quality due to the influence of illumination conditions,topographic changes, and atmospheric effects. Spectral disturbance is corrected on the basis of LMM by introducing the spectral correction term, and a linear mixing model for spectral perturbation correction is proposed. Moreover, an improved HCS method based on modified LMM is proposed. This proposed model only performs spectral compressed sampling on the original hyperspectral images at the sampling end. The proposed method uses the proposed spectral perturbation correction model to reconstruct the original hyperspectral images based on the compressed sampling data. The alternating direction multiplier method is used to estimate the optimal values of each component in the modified LMM to obtain the optimal reconstruction quality. Experimental results show that the proposed method can achieve better reconstructed performance compared with other classical HCS methods.

关 键 词：遥感 高光谱遥感 压缩感知 线性混合模型 光谱扰动 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O433[自动化与计算机技术—计算机科学与技术]