作者：蒋永丛1, 何飞2
作者单位：1. 河南林业职业学院信息与艺术设计系，河南 洛阳 471002;
2. 郑州大学信息工程学院，河南 郑州 450001
关键词：高光谱图像;图拉普拉斯;重加权稀疏;深度非负矩阵;高光谱图像解混
摘要：
为有效挖掘高光谱图像隐层信息以提升混合像元的分解精度，提出一种图正则的重加权稀疏深度非负矩阵分解算法。算法考虑图像丰度矩阵在局部所具有的浅层图结构信息以及在全局所具有的稀疏特性，通过融合图正则项和基于重加权的稀疏正则项及非负矩阵的多层深度结构来提升对混合像元的分解能力。通过乘法更新规则对深度非负矩阵分解算法进行逐层更新以优化全局框架。基于光谱角度距离和均方根误差评价指标，在模拟数据集和真实数据集上的实验显示所提出的算法相比其他典型算法分别有最大约63%和9.7%的解混精度增益。实验证明所提出的图正则重加权稀疏约束的深度非负矩阵分解算法能有效提升高光谱图像的解混精度，更好地服务于国家重大需求。
Graph regularized reweighted sparsity constrained deep nonnegative matrix factorization for hyperspectral image unmixing
JIANG Yongcong1, HE Fei2
1. Department of information and art design, Henan Forestry Vocational College, Henan Luoyang 471002, China;
2. School of Information and Engineering, Zhengzhou University, Zhengzhou, 450001, China
Abstract: In order to effectively exploit the hidden layer information of hyperspectral images and improve the decomposition accuracy of mixed pixels, this paper proposes a graph regularized reweighted sparse deep nonnegative matrix factorization algorithm. The algorithm respectively considers the graph structure information and sparsity properties of the abundance matrix from local and global perspectives. It improves the decomposition ability of mixed pixels by fusing the graph and the reweighted sparse constraints and the multi-layer deep structure of nonnegative matrix. The algorithm is solved layer by layer by multiplication update rules for the purpose of optimizing the global framework. Based on spectral angle distance and root mean square error metrics, experiments on simulated and real data sets show that the proposed algorithm has the largest unmixing accuracy gain of about 63% and 9.7%, respectively, compared with other representative algorithms. Experiments indicate that the proposed algorithm can effectively improve the interpretation accuracy of hyperspectral images and serve the major national needs.
Keywords: hyperspectral image;graph laplacian;reweighted sparsity;deep nonnegative matrix factorization;hyperspectral image unmixing.
2022, 48(12):154-161,180 收稿日期: 2021-10-06;收到修改稿日期: 2022-01-21
基金项目: 国家自然科学基金资助项目(61572444)；河南省青年骨干教师项目(2019GZGG023)
作者简介: 蒋永丛(1983-),男,讲师,主研方向为计算机应用、教育信息化
参考文献
[1] SOWMYA V, SOMAN K P, HASSABALLAH M. Hyperspectral image: fundamentals and advances[M]//Recent Advances in Computer Vision. Springer, Cham, 2019: 401-424
[2] 王小军. 基于PSO优选参数的SVR水质参数遥感反演模型[J]. 中国测试, 2011, 37(1): 66-69
[3] 田文忠, 赵庆展, 胡浩伟, 等. 无人机高光谱载荷性能交叉验证[J]. 中国测试, 2019, 45(11): 131-137
[4] BIOUCAS D M, PLAZA A, et al. Hyperspectral unmixing overview: Geometrical, statistical, and sparse regression-based approaches[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012, 5(2): 354-379
[5] SHI C, WANG L. Incorporating spatial information in spectral unmixing: A review[J]. Remote Sensing of Environment, 2014, 149: 70-87
[6] GHAMISI P, YOKOYA N, LI J, et al. Advances in hyperspectral image and signal processing: A comprehensive overview of the state of the art[J]. IEEE Geoscience and Remote Sensing Magazine, 2017, 5(4): 37-78
[7] NASCIMENTO J M P, BIOUCAS-DIAS J M. Vertex component analysis: A fast algorithm to unmix hyperspectral data[J]. IEEE transactions on Geoscience and Remote Sensing, 2005, 43(4): 898-910
[8] SHEN X, BAO W, QU K. Spatial-spectral hyperspectral endmember extraction using a spatial energy prior constrained maximum simplex volume approach[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 1347-1361
[9] SHEN X, BAO W. Hyperspectral endmember extraction using spatially weighted simplex strategy[J]. Remote Sensing, 2019, 11(18): 2147
[10] ZHANG S, LI J, LI H C, et al. Spectral–spatial weighted sparse regression for hyperspectral image unmixing[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(6): 3265-3276
[11] QI L, LI J, WANG Y, et al. Spectral–spatial-weighted multiview collaborative sparse unmixing for hyperspectral images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(12): 8766-8779
[12] QIAN Y, JIA S, ZHOU J, et al. Hyperspectral unmixing via L1/2 sparsity-constrained nonnegative matrix factorization[J]. IEEE Transactions on Geoscience and Remote Sensing, 2011, 49(11): 4282-4297
[13] WANG X, ZHONG Y, ZHANG L, et al. Spatial group sparsity regularized nonnegative matrix factorization for hyperspectral unmixing[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(11): 6287-6304
[14] RAJABI R, GHASSEMIAN H. "Spectral unmixing of hyperspectral imagery using multilayer NMF. " IEEE Geoscience and Remote Sensing Letters 12.1 (2014): 38-42.
[15] FENG, X, LI H C, LI J, et al. Hyperspectral unmixing using sparsity-constrained deep nonnegative matrix factorization with total variation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(10): 6245-6257
