Xiaoguang Mei

Hyperspectral Image Classification With Robust Sparse Representation

Recently, the sparse representation-based classification (SRC) methods have been successfully used for the classification of hyperspectral imagery, which relies on the underlying assumption that a hyperspectral pixel can be sparsely represented by a linear combination of a few training samples among the whole training dictionary. However, the SRC-based methods ignore the sparse representation residuals (i.e., outliers), which may make the SRC not robust for outliers in practice. To overcome this problem, we propose a robust SRC (RSRC) method which can handle outliers. Moreover, we extend the RSRC to the joint robust sparsity model named JRSRC, where pixels in a small neighborhood around the test pixel are simultaneously represented by linear combinations of a few training samples and outliers. The JRSRC can also deal with outliers in hyperspectral classification. Experiments on real hyperspectral images demonstrate that the proposed RSC and JRSRC have better performances than the orthogonal matching pursuit (OMP) and simultaneous OMP, respectively. Moreover, the JRSRC outperforms some other popular classifiers.

Hyperspectral image denoising using the robust low-rank tensor recovery.

Denoising is an important preprocessing step to further analyze the hyperspectral image (HSI), and many denoising methods have been used for the denoising of the HSI data cube. However, the traditional denoising methods are sensitive to outliers and non-Gaussian noise. In this paper, by utilizing the underlying low-rank tensor property of the clean HSI data and the sparsity property of the outliers and non-Gaussian noise, we propose a new model based on the robust low-rank tensor recovery, which can preserve the global structure of HSI and simultaneously remove the outliers and different types of noise: Gaussian noise, impulse noise, dead lines, and so on. The proposed model can be solved by the inexact augmented Lagrangian method, and experiments on simulated and real hyperspectral images demonstrate that the proposed method is efficient for HSI denoising.

Hyperspectral Unmixing with Robust Collaborative Sparse Regression

Recently, sparse unmixing (SU) of hyperspectral data has received particular attention for analyzing remote sensing images. However, most SU methods are based on the commonly admitted linear mixing model (LMM), which ignores the possible nonlinear effects (i.e., nonlinearity). In this paper, we propose a new method named robust collaborative sparse regression (RCSR) based on the robust LMM (rLMM) for hyperspectral unmixing. The rLMM takes the nonlinearity into consideration, and the nonlinearity is merely treated as outlier, which has the underlying sparse property. The RCSR simultaneously takes the collaborative sparse property of the abundance and sparsely distributed additive property of the outlier into consideration, which can be formed as a robust joint sparse regression problem. The inexact augmented Lagrangian method (IALM) is used to optimize the proposed RCSR. The qualitative and quantitative experiments on synthetic datasets and real hyperspectral images demonstrate that the proposed RCSR is efficient for solving the hyperspectral SU problem compared with the other four state-of-the-art algorithms.

An Infrared Small Target Detecting Algorithm Based on Human Visual System

Infrared (IR) small target detection with high detection rate, low false alarm rate, and multiscale detection ability is a challenging task since raw IR images usually have low contrast and complex background. In recent years, robust human visual system (HVS) properties have been introduced into the IR small target detection field. However, existing algorithms based on HVS, such as difference of Gaussians (DoG) filters, are sensitive to not only real small targets but also background edges, which results in a high false alarm rate. In this letter, the difference of Gabor (DoGb) filters is proposed and improved (IDoGb), which is an extension of DoG but is sensitive to orientations and can better suppress the complex background edges, then achieves a lower false alarm rate. In addition, multiscale detection can be also achieved. Experimental results show that the IDoGb filter produces less false alarms at the same detection rate, while consuming only about 0.1 s for a single frame.

GBM-Based Unmixing of Hyperspectral Data Using Bound Projected Optimal Gradient Method

The generalized bilinear model (GBM) has been widely used for the nonlinear unmixing of hyperspectral images, and traditional GBM solvers include the Bayesian algorithm, the gradient descent algorithm, the semi-nonnegative-matrix-factorization algorithm, etc. However, they suffer from one of the following problems: high computational cost, sensitive to initialization, and the pixelwise algorithm hinders us from applying to large hyperspectral images. In this letter, we apply Nesterovs optimal gradient method to solve the least-square problem under the bound constraint, which is named as the bound projected optimal gradient method (BPOGM). The BPOGM can achieve the optimal convergence rate of O(1/k2), with k denoting the number of iterations in BPOGM. We further apply the BPOGM to solve the GBM-based unmixing problem. Experiments on both synthetic data sets and real hyperspectral images demonstrate that the BPOGM is efficient for solving the GBM-based unmixing problem.

Hyperspectral image denoising with superpixel segmentation and low-rank representation

Recently, low-rank representation (LRR) based hyperspectral image (HSI) restoration method has been proven to be a powerful tool for simultaneously removing different types of noise, such as Gaussian, dead pixels and impulse noise. However, the LRR based method just adopts the square patch denoising strategy, which makes it not able to excavate the spatial information in HSI. This paper integrates superpixel segmentation (SS) into LRR and proposes a novel denoising method called SSLRR. First, the principal component analysis (PCA) is adopted to obtain the first principal component of HSI. Then the SS is adopted to the first principal component of HSI to get the homogeneous regions. Since we excavate the spatial-spectral information of HSI by combining PCA with SS, it is better than simply dividing the HSI into square patches. Finally, we employ the LRR to each homogeneous region of HSI, which enable us to remove all the above mentioned different types of noise simultaneously. Extensive experiments conducted on synthetic and real HSIs indicate that the SSLRR is efficient for HSI denoising.

Sparse Unmixing of Hyperspectral Data with Noise Level Estimation

Recently, sparse unmixing has received particular attention in the analysis of hyperspectral images (HSIs). However, traditional sparse unmixing ignores the different noise levels in different bands of HSIs, making such methods sensitive to different noise levels. To overcome this problem, the noise levels at different bands are assumed to be different in this paper, and a general sparse unmixing method based on noise level estimation (SU-NLE) under the sparse regression framework is proposed. First, the noise in each band is estimated on the basis of the multiple regression theory in hyperspectral applications, given that neighboring spectral bands are usually highly correlated. Second, the noise weighting matrix can be obtained from the estimated noise. Third, the noise weighting matrix is integrated into the sparse regression unmixing framework, which can alleviate the impact of different noise levels at different bands. Finally, the proposed SU-NLE is solved by the alternative direction method of multipliers. Experiments on synthetic datasets show that the signal-to-reconstruction error of the proposed SU-NLE is considerably higher than those of the corresponding traditional sparse regression unmixing methods without noise level estimation, which demonstrates the efficiency of integrating noise level estimation into the sparse regression unmixing framework. The proposed SU-NLE also shows promising results in real HSIs.

Infrared ultraspectral signature classification based on a restricted Boltzmann machine with sparse and prior constraints

The state-of-the-art ultraspectral technology brings a new hope for the high precision applications due to its high spectral resolution. However, it comes with new challenges brought by the improvement of spectral resolution such as the Hughes phenomenon and over-fitting issue, and our work is aimed at addressing these problems. As new Markov random field (MRF) models, the restricted Boltzmann machines (RBMs) have been used as generative models for many different pattern recognition and artificial intelligence applications showing promising and outstanding performance. In this article, we propose a new method for infrared ultraspectral signature classification based on the RBMs, which adopt the regularization-based techniques to improve the classification accuracy and robustness to noise compared to traditional RBMs. First, we add an arctan-like term to the objective function as a sparse constraint to improve the classification accuracy. Second, we utilize a Gaussian prior to avoid the over-fitting problem. Third, to further improve the classification performance, a multi-layer RBM model, a deep belief network (DBN), is adopted for infrared ultraspectral signature classification. Experiments using different spectral libraries provided by the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and the Environmental Protection Agency (EPA) were performed to evaluate the performance of the proposed method by comparing it with other traditional methods, including spectral coding-based classifiers (binary coding (BC), spectral feature-based binary coding (SFBC), and spectral derivative feature coding (SDFC) matching methods), a novel feature extraction method termed crosscut feature extraction matching (CF), and three machine learning methods (artificial deoxyribonucleic acid (DNA)-based spectral matching (ADSM), DBN, and sparse deep belief network (SparseDBN)). Experimental results demonstrate that the proposed method is superior to the other methods with which it was compared and can simultaneously improve the accuracy and robustness of classification.

Robust GBM hyperspectral image unmixing with superpixel segmentation based low rank and sparse representation

Abstract Generalized bilinear model (GBM) has been one of the most representative models for nonlinear unmixing of hyperspectral image (HSI), which can take the second-order scattering of photons into consideration. However, the GBM is implicitly developed for the additive white Gaussian noise. Besides, the performances of traditional GBM based unmixing methods are not that satisfying since the spatial correlation of HSI is not considered. In this paper, to overcome the two problems mentioned above, we propose a robust GBM (RGBM) for nonlinear unmixing of HSI, which can simultaneously take the Gaussian noise and sparse noise into account. Besides, we propose a new unmixing method with superpixel segmentation (SS) and low-rank representation (LRR) based on RGBM, which can take the spatial correlation of HSI into consideration. First, we adopt the principal component analysis (PCA) to get the first principal component of HSI, which contains the most information for the whole HSI. Then we adopt the SS in the first principal component of HSI to get the homogeneous regions, and the abundances in each homogeneous region have the underlying low-rank property. Finally, we unmix the pixels in each homogeneous region of HSI according to the low-rank property of abundances and the sparse property of sparse noise, and the proposed RGBM based unmixing method can be solved by the alternative direction method of multipliers (ADMM). Experiments on both synthetic datasets and real HSIs demonstrate that the proposed RGBM and corresponding method are efficient compared with some other popular GBM based unmixing methods.

Robust Image Feature Matching via Progressive Sparse Spatial Consensus

In this paper, we propose an efficient algorithm, termed as progressive sparse spatial consensus, for mismatch removal from a set of putative feature correspondences involving large number of outliers. Our goal is to estimate the underlying spatial consensus between the feature correspondences and then remove mismatches accordingly. This is formulated as a maximum likelihood estimation problem, and solved by an iterative expectation-maximization algorithm. To handle large number of outliers, we introduce a progressive framework, which uses matching results on a small putative set with high inlier ratio to guide the matching on a large putative set. The spatial consensus is modeled by a non-parametric thin-plate spline kernel; this enables our method to handle image pairs with both rigid and non-rigid motions. Moreover, we also introduce a sparse approximation to accelerate the optimization, which can largely reduce the computational complexity without degenerating the accuracy. The quantitative results on various experimental data demonstrate that our method can achieve better matching accuracy and can generate more good matches compared to several state-of-the-art methods.