Hypergraph Regularized Deep Autoencoder for Unsupervised Unmixing Hyperspectral Images

Deep learning(DL)has shown its superior performance in dealing with various computer vision tasks in recent years.As a simple and effective DL model,autoencoder(AE)is popularly used to decompose hyperspectral images(HSIs)due to its powerful ability of feature extraction and data reconstruction.However,most existing AE-based unmixing algorithms usually ignore the spatial information of HSIs.To solve this problem,a hypergraph regularized deep autoencoder(HGAE)is proposed for unmixing.Firstly,the traditional AE architecture is specifically improved as an unsupervised unmixing framework.Secondly,hypergraph learning is employed to reformulate the loss function,which facilitates the expression of high-order similarity among locally neighboring pixels and promotes the consistency of their abundances.Moreover,L_(1/2)norm is further used to enhance abundances sparsity.Finally,the experiments on simulated data,real hyperspectral remote sensing images,and textile cloth images are used to verify that the proposed method can perform better than several state-of-the-art unmixing algorithms.

Recursive Dictionary-Based Simultaneous Orthogonal Matching Pursuit for Sparse Unmixing of Hyperspectral Data

The sparse unmixing problem of greedy algorithms still remains a great challenge at finding an optimal subset of endmembers for the observed data from the spectral library,due to the usually high correlation of the spectral library.Under such circumstances,a novel greedy algorithm for sparse unmixing of hyperspectral data is presented,termed the recursive dictionary-based simultaneous orthogonal matching pursuit(RD-SOMP).The algorithm adopts a block-processing strategy to divide the whole hyperspectral image into several blocks.At each iteration of the block,the spectral library is projected into the orthogonal subspace and renormalized,which can reduce the correlation of the spectral library.Then RD-SOMP selects a new endmember with the maximum correlation between the current residual and the orthogonal subspace of the spectral library.The endmembers picked in all the blocks are associated as the endmember sets of the whole hyperspectral data.Finally,the abundances are estimated using the whole hyperspectral data with the obtained endmember sets.It can be proved that RD-SOMP can recover the optimal endmembers from the spectral library under certain conditions.Experimental results demonstrate that the RD-SOMP algorithm outperforms the other algorithms,with a better spectral unmixing accuracy.

Multiple Endmember Hyperspectral Sparse Unmixing Based on Improved OMP Algorithm

In conventional linear spectral mixture analysis model,a class is represented by a single endmember.However,the intra-class spectral variability is usually very large,which makes it difficult to represent a class,and in this case,it leads to incorrect unmixing results. Some proposed algorithms play a positive role in overcoming the endmember variability,but there are shortcomings on computation intensive,unsatisfactory unmixing results and so on. Recently,sparse regression has been applied to unmixing,assuming each mixed pixel can be expressed as a linear combination of only a few spectra in a spectral library. It is essentially the same as multiple endmember spectral unmixing. OMP( orthogonal matching pursuit),a sparse reconstruction algorithm,has advantages of simple structure and high efficiency. However,it does not take into account the constraints of abundance non-negativity and abundance sum-to-one( ANC and ASC),leading to undesirable unmixing results. In order to solve these issues,this paper presents an improved OMP algorithm( fully constraint OMP,FOMP) for multiple endmember hyperspectral sparse unmixing. The proposed algorithm overcomes the shortcomings of OMP,and on the other hand,it solves the problem of endmember variability.The ANC and ASC constraints are firstly added into the OMP algorithm,and then the endmember set is refined by the relative increase in root-mean-square-error( RMSE) to avoid over-fitting,finally pixels are unmixed by their optimal endmember set. The simulated and real hyperspectral data experiments show that FOPM unmixing results are ideally comparable and abundance RMSE reduces much lower than OMP and simple spectral mixture analysis( s SMA),and has a strong anti-noise performance. It proves that multiple endmember spectral mixture analysis is more reasonable.

Unsupervised hyperspectral unmixing based on robust nonnegative dictionary learning

Considering the sparsity of hyperspectral images(HSIs),dictionary learning frameworks have been widely used in the field of unsupervised spectral unmixing.However,it is worth mentioning here that existing dictionary learning method-based unmixing methods are found to be short of robustness in noisy contexts.To improve the performance,this study specifically puts forward a new unsupervised spectral unmixing solution.For the reason that the solution only functions in a condition that both endmembers and the abundances meet non-negative con-straints,a model is built to solve the unsupervised spectral un-mixing problem on the account of the dictionary learning me-thod.To raise the screening accuracy of final members,a new form of the target function is introduced into dictionary learning practice,which is conducive to the growing robustness of noisy HSI statistics.Then,by introducing the total variation(TV)terms into the proposed spectral unmixing based on robust nonnega-tive dictionary learning(RNDLSU),the context information under HSI space is to be cited as prior knowledge to compute the abundances when performing sparse unmixing operations.Ac-cording to the final results of the experiment,this method makes favorable performance under varying noise conditions,which is especially true under low signal to noise conditions.

Robust Deep 3D Convolutional Autoencoder for Hyperspectral Unmixing with Hypergraph Learning

Hyperspectral unmixing aims to acquire pure spectra of distinct substances(endmembers)and fractional abundances from highly mixed pixels.In this paper,a deep unmixing network framework is designed to deal with the noise disturbance.It contains two parts:a three⁃dimensional convolutional autoencoder(denoising 3D CAE)which recovers data from noised input,and a restrictive non⁃negative sparse autoencoder(NNSAE)which incorporates a hypergraph regularizer as well as a l2,1⁃norm sparsity constraint to improve the unmixing performance.The deep denoising 3D CAE network was constructed for noisy data retrieval,and had strong capacity of extracting the principle and robust local features in spatial and spectral domains efficiently by training with corrupted data.Furthermore,a part⁃based nonnegative sparse autoencoder with l2,1⁃norm penalty was concatenated,and a hypergraph regularizer was designed elaborately to represent similarity of neighboring pixels in spatial dimensions.Comparative experiments were conducted on synthetic and real⁃world data,which both demonstrate the effectiveness and robustness of the proposed network.

CUR Based Initialization Strategy for Non-Negative Matrix Factorization in Application to Hyperspectral Unmixing

Hyperspectral unmixing is a powerful tool for the remote sensing image mining. Nonnegative matrix factorization (NMF) has been adopted to deal with this issue, while the precision of unmixing is closely related with the local minimizers of NMF. We present two novel initialization strategies that is based on CUR decomposition, which is physically meaningful. In the experimental test, NMF with the new initialization method is used to unmix the urban scene which was captured by airborne visible/infrared imaging spectrometer (AVIRIS) in 1997, numerical results show that the initialization methods work well.

A Novel Fuzzy Inference System-Based Endmember Extraction in Hyperspectral Images

Spectral unmixing helps to identify different components present in the spectral mixtures which occur in the uppermost layer of the area owing to the low spatial resolution of hyperspectral images.Most spectral unmixing methods are globally based and do not consider the spectral variability among its endmembers that occur due to illumination,atmospheric,and environmental conditions.Here,endmember bundle extraction plays a major role in overcoming the above-mentioned limitations leading to more accurate abundance fractions.Accordingly,a two-stage approach is proposed to extract endmembers through endmember bundles in hyperspectral images.The divide and conquer method is applied as the first step in subset images with only the non-redundant bands to extract endmembers using the Vertex Component Analysis(VCA)and N-FINDR algorithms.A fuzzy rule-based inference system utilizing spectral matching parameters is proposed in the second step to categorize endmembers.The endmember with the minimum error is chosen as the final endmember in each specific category.The proposed method is simple and automatically considers endmember variability in hyperspectral images.The efficiency of the proposed method is evaluated using two real hyperspectral datasets.The average spectral angle and abundance angle are used to analyze the performance measures.

Hyperspectral Image Sharpening Based on Deep Convolutional Neural Network and Spatial-Spectral Spread Transform Models

In order to improve the spatial resolution of hyperspectral(HS)image and minimize the spectral distortion,an HS and multispectral(MS)image fusion approach based on convolutional neural network(CNN)is proposed.The proposed approach incorporates the linear spectral mixture model and spatial-spectral spread transform model into the learning phase of network,aiming to fully exploit the spatial-spectral information of HS and MS images,and improve the spectral fidelity of fusion images.Experiments on two real remote sensing data under different resolutions demonstrate that compared with some state-of-the-art HS and MS image fusion methods,the proposed approach achieves superior spectral fidelities and lower fusion errors.

基于光易变性低秩正交先验的高光谱解混

高光谱解混是通过图像分解提取端元及丰度特征的过程,然而由光照、大气等因素引起的光谱类内易变性,或者由环境变化、设备等非线性因素导致的谱间易变性,会导致特征提取精度下降。为了全面考虑解混过程中光谱变化的问题,本文引入光谱易变性的低秩正交先验提出了一种增强型的光谱解混优化模型。首先,在线性解混模型基础上引入易变性数据拟合项来同时考虑光谱类内和类间变化,利用缩放因子来解决光谱类内易变性,同时增加光谱易变性扰动矩阵来解决谱间易变性。其次,该模型利用正交先验约束来实现原光谱字典与易变性项的空间低相干性,通过采用核范数对数松弛来强化丰度矩阵的低秩特性,抑制微小分量及噪声。最后,采用交替优化法及向量-矩阵算子降低求解算法复杂度。通过模拟数据集和真实数据集仿真测试结果表明,本文所提算法取得了优于对比算法的良好性能,验证了该优化模型的有效性。

同质区共享端元变异性的高光谱混合像元分解

由于不同的照明条件、复杂的大气环境等因素,相同端元的光谱特征在图像的不同位置呈现出可见的差异,这种现象被称为端元的光谱变异性。在相当大的场景中,端元的变异性可能很大,但在适度的局部同质区内,变异性往往很小。扰动线性混合模型(Perturbed Linear Mixing Model,PLMM)在解混的过程中可以减轻端元变异性造成的不利影响,但是对缩放效应造成的变异性的处理能力较弱。为此,本文改进了扰动线性混合模型,引入了尺度因子以处理缩放效应造成的变异性,并结合超像素分割算法划分局部同质区,然后设计出基于局部同质区共享端元变异性的解混算法(Shared Endmember Variability in Unmixing,SEVU)。与扰动线性混合模型,扩展线性混合模型(Extended Linear Mixing Model,ELMM)等算法相比,所提SEVU算法在合成数据集上平均端元光谱角距离(mean Spectral Angle Distance,mSAD)和丰度均方根误差(abundance Root Mean Square Error,aRMSE)最优,分别为0.0855和0.0562;在Jasper Ridge和Cuprite真实数据集上mSAD是最优的,分别为0.0603和0.1003。在合成数据集和两个实测数据集上的实验结果验证了SEVU算法的有效性。

基于光谱降维与Hu矩的壁画颜料层脱落区域提取方法

颜料层脱落区域的提取是壁画现状调查的重要环节,由于其光谱特征与壁画白色图案较为相似,仅利用光谱特征提取的精度较低。因此,提出了一种兼顾光谱特征和Hu矩形状特征的颜料层脱落区域提取方法。首先,利用壁画高光谱图像的光谱信息,经光谱降维,采用支持向量机监督分类法提取颜料层脱落区域与白色图案。然后,对颜料层脱落区域与白色图案分类结果分别进行连通,将连通后的图斑视为最小识别对象,利用Hu矩计算每一个对象的形状特征,采用支持向量机二分类再次区分对象图斑,实现颜料层脱落区域的半自动提取。最后,以青海省瞿昙寺壁画高光谱图像进行了提取。结果表明,该方法能提高颜料层脱落区域的提取精度,为壁画的现状调查提供支撑。

Area-Correlated Spectral Unmixing Based on Bayesian Nonnegative Matrix Factorization

To solve the problem of the spatial correlation for adjacent areas in traditional spectral unmixing methods, we propose an area-correlated spectral unmixing method based on Bayesian nonnegative matrix factorization. In the proposed me-thod, the spatial correlation property between two adjacent areas is expressed by a priori probability density function, and the endmembers extracted from one of the adjacent areas are used to estimate the priori probability density func-tions of the endmembers in the current area, which works as a type of constraint in the iterative spectral unmixing process. Experimental results demonstrate the effectivity and efficiency of the proposed method both for synthetic and real hyperspectral images, and it can provide a useful tool for spatial correlation and comparation analysis between ad-jacent or similar areas.

利用光谱解混合的目标检测

高光谱目标检测中背景信息的统计往往受到目标信息的干扰,而高光谱图像中存在的大量混合像元会进一步加深这一干扰。为了准确统计背景信息、显著降低目标像元对背景统计信息的干扰,提出了一种利用光谱解混合的目标检测算法,通过光谱解混合和目标相似性判断,获取目标端元对应丰度系数,并与光谱夹角系数相结合生成合理的背景加权系数,进行加权约束最小能量算子(CEM)目标检测,从而有效提高混合像元的背景信息统计准确度;利用目标端元对应丰度系数和光谱夹角系数生成初步的目标检测结果,与加权CEM目标检测结果相融合进行进一步优化,有效提高算法稳定性,同时再次提高目标检测精度。实验结果表明:对于模拟高光谱图像和真实高光谱图像,本文算法均得到了较好的目标检测效果,算法稳定性较强,且有效提高了目标检测精度,相比传统CEM算法、基于光谱角的加权CEM算法、归一化丰度系数作为目标结果,AUC值分别平均提高了0.0712,0.0312和0.0150,在高光谱应用中具有较强的实用性。

双约束深度卷积网络的高光谱图像空谱解混方法

高光谱图像凭借其“图谱合一”的特点逐渐在军事、环境、农业等方面发挥出重要作用。但是,由于传感器空间分辨率的限制以及地物分布的复杂多样性,高光谱遥感图像中通常存在大量的混合像元,严重制约了高光谱遥感的应用范围。目前,处理混合像元问题最有效的分析方法是混合像元分解(解混)。近年来,深度学习的发展对高光谱遥感产生了重大影响,也催生出一系列基于深度学习的解混方法。现有基于深度学习的解混方法在隐藏信息挖掘方面表现出极大的潜力和优势,通常情况下能够取得更加准确的结果。然而,这些方法大多只考虑了地物的光谱信息而忽略空间分布规律,导致在复杂场景中估算结果可能并不理想,逐渐难以满足工程应用的实际需求。为进一步发掘和利用空间信息提升解混的准确性,本文构建了一种新的深度学习网络来实现高光谱图像解混。新提出的解混网络采用卷积层来获取先验信息,利用高斯核函数的特性来协助区分物质属性,并且通过分配中心像元与邻域像元间的权重来增进丰度平滑性。在新网络中,本文使用Softmax作为丰度对应层的激活函数来约束丰度的输出。此外,在Softmax中,本文采用了L1/2正则化来避免节点出现过拟合而影响最终结果,进一步强化了网络性能,最终形成了一种双约束强化的深度卷积自编码网络来实现无监督的解混。为了验证新方法的有效性和优势,本文将新提出的方法与同类解混方法应用在一系列高光谱数据(包括模拟图像和真实图像)中进行测试,均达到了预期效果。本文的研究成果能够为处理混合像元问题提供了新的技术支撑和理论参考。

基于深度自编码网络的高光谱影像解混研究

高分系列卫星的发射和无人机高光谱技术的发展,高光谱可用数据进一步扩展。为了提升高光谱数据的精细利用价值,高光谱影像混合像元解混成为当前至关重要的任务。随着人工智能技术的快速发展,深度学习理论被引入遥感图像处理领域。自编码网络具有较强的特征提取能力,已经开始应用于高光谱影像解混方面。以自编码网络为基础对其结构进行改进,提出一种深度堆栈自编码网络(DSAE)用于高光谱图像解混研究。该网络包含两个部分:端元识别网络(EDSAE)和丰度求解的网络(ADSAE)。首先,通过添加批标准化处理、稀疏约束、“和为一”约束以及删除网络偏置项构建EDSAE网络,开展非监督训练进行高光谱影像端元识别。其次,将获取的端元光谱数据依据HAPKE非线性混合模型和LINEAR线性混合模型开展数据增强,生成多元混合的带有丰度标签的模拟高光谱数据集。最后,在堆栈自编码网络基础上,设置最后一层自编码器的激活函数为Softmax函数,构建监督训练网络ADSAE,把模拟数据集作为训练数据,高光谱影像作为测试数据,求取真实高光谱影像的丰度矩阵。对Samson、 Jasper Ridge和Urban公共的高光谱影像开展端元识别和丰度求解实验,基于DSAE获得的结果与传统的N-FINDR、 VCA、 MVC-NMF方法以及目前已有深度学习的方法SNSA和EndNet取得的结果进行比较。结果表明:对3组真实的高光谱影像开展解混,DSAE方法在端元提取方面相比于其他5种方法,具有最优精度;在丰度求解方面,基于HAPKE模型生成的模拟数据集,利用ADSAE网络开展监督训练可以成功获得3组高光谱影像的丰度矩阵,相比于LINEAR模型和FCLS方法,均具有最优的丰度反演结果。DSAE方法具有较好的稳定性和鲁棒性,为高光谱影像定量研究提供了新的思路。

高光谱遥感图像亚像元信息提取方法综述

高光谱遥感图像光谱分辨率高、波谱连续、图谱合一,这为精细地物分类、探测和识别提供了数据基础。然而,由于高光谱遥感图像空间分辨率的局限性及地物场景的复杂分布,混合像元普遍存在于高光谱遥感图像。混合像元是高光谱遥感图像精细信息提取与分析中的难点。解决混合像元问题,实现亚像元级信息的提取与分析是近年来高光谱遥感图像解译的热点和前沿。本文系统梳理了高光谱遥感图像亚像元信息提取的主要研究内容,具体从混合像元分解、亚像元制图及亚像元目标探测3个研究方向综述了经典方法,并对国内外相关方向的研究进展、发展前沿及主要挑战进行了分析与评价,最后分析讨论了高光谱遥感图像亚像元信息提取研究在模型构建、优化求解及与应用结合等方面的研究趋势及方向。

多端元模式下高光谱图像解混的不确定性问题

为了探究多端元解混造成解混结果不确定性的原因,并进一步在解混中克服与降低该不确定性,详细分析了在二分类解混中,解混不确定性的两种表现形式,丰度固定时像元位置的不确定性和像元位置固定时丰度的不确定性,探究解混端元与不确定性的相互作用关系,进而提出一种可降低解混丰度不确定性的端元加权多端元解混方法。实验表明:混合丰度不确定性的存在,同时在保证解混精度的前提下验证了所提出的降低不确定性方法的有效性。

全变差稀疏约束深度非负矩阵分解高光谱遥感影像解混方法

传统非负矩阵分解方法仅基于单层线性模型,现有的深度非负矩阵分解模型忽略了地物光谱的实际混合物理过程,仅从数学理论考虑深度分解。对此,文中从光谱混合的物理过程出发,综合非负矩阵分解和深度学习,将光谱混合过程进行反向建模,并充分考虑丰度的稀疏性和空间平滑性,构建了用于高光谱遥感影像解混的面向端元矩阵的全变差稀疏约束深度非负矩阵分解模型。通过模拟实验和真实实验,将文中所提方法与5种解混方法进行对比。结果表明,相较于面向丰度的深度非负矩阵分解算法,文中所提方法的平均光谱角距离和均方根误差均有所降低,取得了最佳解混结果。

基于多模态多目标优化的端元束提取方法研究

为解决高光谱影像受传感器及分辨率的影响所产生的光谱变化给解混造成的困扰,提出基于多模态多目标优化的端元束提取方法(MOPSOSCD)。对高光谱图像进行标号编码,采用基于索引的环形拓扑结构进行邻域的个体交互,通过邻域最优改进粒子群速度更新方式并整数化粒子位置更新。同时,根据高光谱图像空间特征,通过改进决策空间拥挤距离提高决策空间的多样性,再结合目标空间的拥挤距离进行综合排序,实现多模态多目标优化的粒子筛选。当粒子定向移动概率pm为0.2、粒子数P为30及迭代次数M为400时,算法在MUUFL数据集上均方根误差(RMSE)及平均光谱角距离(mSAD)分别为0.0088、0.1112。通过对比试验,本文方法相较于VCA、DPSO等方法具有更高的提取精度和效率,为高光谱解混提供了更加准确的端元束提取方法。