基于Hessian图正则稀疏NMF的高光谱解混
    点此下载全文
引用本文:汤辉,孟莎莎,彭天亮,付康.基于Hessian图正则稀疏NMF的高光谱解混[J].计算技术与自动化,2023,(1):153-159
摘要点击次数: 196
全文下载次数: 0
作者单位
汤辉,孟莎莎,彭天亮,付康 (江西省科技基础条件平台中心江西 南昌 330002) 
中文摘要:基于非负矩阵分解(Nonnegative Matrix Factorization, NMF)的高光谱解混(Hyperspectral Unmixing ,HU)方法引起了大家的关注,因为可以将一个非负高光谱图像(Hyperspectral Imagery,HSI)数据矩阵分解为两个非负矩阵的乘积,分别对应于端元矩阵和丰度系数矩阵。目前,图约束的NMF算法已经被证明对高光谱解混是有效的,因为它们可以捕获HSI的几何特性。为了挖掘数据在混合过程中的几何结构和稀疏性,提出了一种稀疏的Hessian图正则化NMF(SHGNMF)算法。SHGNMF算法是将丰度矩阵的L1/2正则化器和Hessian图正则化项都添加到每个NMF模型中,同时采用乘法更新规则。最后用模拟数据和真实数据进行实验,验证了所提出的SHGNMF算法相对于其他NMF算法的优越性。
中文关键词:高光谱解混  NMF  稀疏  Hessian图正则化  高光谱图像
 
Hyperspectral Unmixing Based on Hessian Graph Regular Sparse NMF
Abstract:Hyperspectral unmixing (HU) method based on nonnegative matrix factorization (NMF) has attracted much attention because it can decompose a hyperspectral image (HSI) data matrix into the product of two nonnegative matrices, corresponding to the end element matrix and the abundance coefficient matrix respectively. At present, graph constrained NMF algorithms have been proved to be effective for hyperspectral unmixing because they can capture the geometric characteristics of HSI. In order to mine the geometric structure and sparsity of data in the mixing process, a sparse Hessian graph regularization NMF (SHGNMF) algorithm is proposed in this paper. SHGNMF algorithm adds the L1/2 regularizer of abundance matrix and the regularization term of Hessian graph to each NMF model, and adopts the multiplication update rule at the same time. Finally, experiments with simulated and real data verify the superiority of the proposed SHGNMF algorithm over other NMF algorithms.
keywords:hyperspectral unmixing  NMF  sparsity  Hessian graph regularization  hyperspectral image
查看全文   查看/发表评论   下载pdf阅读器