一种低秩和图正则化的协同稀疏高光谱解混方法  

A Low-rank and Graph Regularization Collaborative Sparse Hyperspectral Unmixing Method

作　　者：韩红伟 陈聆[2] 苗加庆 HAN Hongwei;CHEN Ling;MIAO Jiaqing(Department of Basic Education,The Engineering&Technical College of Chengdu University of Technology,Leshan 614000,China;Geomathematics Key Laboratory of Sichuan Province,Chengdu University of Technology,Chengdu 610059,China;School of Mathematics,Southwest Minzu University,Chengdu 610041,China)

机构地区：[1]成都理工大学工程技术学院基础教学部,四川乐山614000 [2]成都理工大学数学地质四川省重点实验室,四川成都610059 [3]西南民族大学数学学院,四川成都610041

出　　处：《无线电工程》2023年第4期868-876,共9页Radio Engineering

基　　金：四川省科技厅-中央引导地方项目(2021ZYD0021);数学地质四川省重点实验室开放基金项目(scsxdz2021zd01,scsxdz2019yb01)。

摘　　要：针对经典协同稀疏解混方法中稀疏性表征不足以及丰度矩阵过平滑等问题,提出一种低秩和图正则化的协同稀疏高光谱解混方法。引入加权因子,进一步促进丰度矩阵的稀疏性;引入了图正则化项,获取图像的空间信息,以促进图像的平滑性;在模型中增加低秩项,进而挖掘高光谱数据的细节结构,进一步提高解混的精度。利用2个模拟和1个真实高光谱数据进行实验,结果表明,提出方法的解混精度与经典解混方法相比得到显著提升。A low-rank and graph regularization collaborative sparse hyperspectral unmixing method is proposed to address the lack of the sparsity of abundance in classical collaborative sparse unmixing methods and the excessive smoothness of abundance matrix.Firstly,a weighted factor is utilized to further promote the sparsity of abundance matrix.Secondly,a graph regularization term is employed to capture the spatial information of the image to promote the smoothness of the image.Finally,a low-rank term is added to the model to explore the detailed structure of hyperspectral data and further improve the accuracy of unmixing.Two simulated hyperspectral data and one real hyperspectral data are used for experiments,and the experimental results show that the proposed algorithm is more accurate than other classical methods.

关 键 词：高光谱图像 稀疏 低秩 光谱解混 图正则化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]