&lt;title&gt;Curvilinear component analysis for nonlinear dimensionality reduction of hyperspectral images&lt;/title&gt;

Lennon, Marc; Mercier, Grégoire; Mouchot, Marie-Catherine; Hubert‐Moy, Laurence

doi:10.1117/12.454150

Cited by 31 publications

(25 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, curvilinear component analysis [26], curvilinear distance analysis [27], and manifold learning [28]- [31] are nonlinear projections based on the preservation of the local topology. Independent component analysis [32], [33], projection pursuit [34], [35], and wavelet decomposition [36], [37] have also been considered.…”

Section: A Unsupervised Hyperspectral Dimensionality Reduction Methodsmentioning

confidence: 99%

Hyperspectral Subspace Identification

Bioucas-Dias

Nascimento

2008

IEEE Trans. Geosci. Remote Sensing

1,008

570

View full text Add to dashboard Cite

Abstract-Signal subspace identification is a crucial first step in many hyperspectral processing algorithms such as target detection, change detection, classification, and unmixing. The identification of this subspace enables a correct dimensionality reduction, yielding gains in algorithm performance and complexity and in data storage. This paper introduces a new minimum mean square error-based approach to infer the signal subspace in hyperspectral imagery. The method, which is termed hyperspectral signal identification by minimum error, is eigen decomposition based, unsupervised, and fully automatic (i.e., it does not depend on any tuning parameters). It first estimates the signal and noise correlation matrices and then selects the subset of eigenvalues that best represents the signal subspace in the least squared error sense. State-of-the-art performance of the proposed method is illustrated by using simulated and real hyperspectral images.Index Terms-Dimensionality reduction, hyperspectral imagery, hyperspectral signal subspace identification by minimum error (HySime), hyperspectral unmixing, linear mixture, minimum mean square error (mse), subspace identification.

show abstract

Section: A Unsupervised Hyperspectral Dimensionality Reduction Methodsmentioning

confidence: 99%

Hyperspectral Subspace Identification

Bioucas-Dias

Nascimento

2008

IEEE Trans. Geosci. Remote Sensing

1,008

570

View full text Add to dashboard Cite

show abstract

“…A variety of local methods exist for estimating manifolds. For example, curvilinear component analysis [100], curvilinear distance analysis [101], manifold learning [102]- [107] are non-linear projections based on the preservation of the local topology. Independent component analysis [108], [109], projection pursuit [110], [111], and wavelet decomposition [112], [113] have also been considered.…”

Section: Signal Subspace Identificationmentioning

confidence: 99%

Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches

Bioucas-Dias

Plaza

Dobigeon

et al. 2012

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

2,304

544

View full text Add to dashboard Cite

International audienceImaging spectrometers measure electromagnetic energy scattered in their instantaneous field view in hundreds or thousands of spectral channels with higher spectral resolution than multispectral cameras. Imaging spectrometers are therefore often referred to as hyperspectral cameras (HSCs). Higher spectral resolution enables material identification via spectroscopic analysis, which facilitates countless applications that require identifying materials in scenarios unsuitable for classical spectroscopic analysis. Due to low spatial resolution of HSCs, microscopic material mixing, and multiple scattering, spectra measured by HSCs are mixtures of spectra of materials in a scene. Thus, accurate estimation requires unmixing. Pixels are assumed to be mixtures of a few materials, called endmembers. Unmixing involves estimating all or some of: the number of endmembers, their spectral signatures, and their abundances at each pixel. Unmixing is a challenging, ill-posed inverse problem because of model inaccuracies, observation noise, environmental conditions, endmember variability, and data set size. Researchers have devised and investigated many models searching for robust, stable, tractable, and accurate unmixing algorithms. This paper presents an overview of unmixing methods from the time of Keshava and Mustard's unmixing tutorial [1] to the present. Mixing models are first discussed. Signal-subspace, geometrical, statistical, sparsity-based, and spatial-contextual unmixing algorithms are described. Mathematical problems and potential solutions are described. Algorithm characteristics are illustrated experimentally

show abstract

“…Even though its theoretical limitations for hyperspectral data analysis have been pointed out (Landgrebe 2003;Lennon 2002), in a practical situation the results obtained using the PCA are still competitive for the purpose of classification (Journaux et al 2006;Lennon et al 2001). The advantages of the PCA are its low complexity and the absence of parameters.…”

Section: Feature Extraction and Selectionmentioning

confidence: 99%

Machine Learning Techniques in Remote Sensing Data Analysis

Waske

Fauvel

Benediktsson

et al. 2009

Kernel Methods for Remote Sensing Data Analysis

View full text Add to dashboard Cite

Several applications have been developed in the field of remote sensing image analysis during the last decades. Besides well-known statistical approaches, many recent methods are based on techniques taken from the field of machine learning. A major aim of machine learning algorithms in remote sensing is supervised classification, which is perhaps the most widely used image classification approach. In this chapter a brief introduction to machine learning and the different paradigms in remote sensing is given. Moreover this chapter briefly discusses the use of recent developments in supervised classification techniques such as neural networks, support vector machines and multiple classifier systems. Introduction Challenges in remote sensingOwing to the recent development of different Earth observation platforms with increased spatial and spectral resolution as well as higher revisit times, remote sensing provides more detailed information on land cover and the environmental state than ever before. Moreover, different Earth-observation systems, such as multi-spectral and SAR systems operate in different wavelengths, ranging from visible to microwave. The data sets consequently provide different, but complementary information. The classification of such data might be considered complex on the one hand, but with regard to recent and upcoming missions, remote sensing applications become even more attractive, on the other.Many early techniques have been taken directly from signal processing and these methods are often based on simple data models and approaches. However, when dealing with recent data sets these well-known classifiers can be limited (Richards 2005). In addition, increased performance requirements such as speed (e.g., for operational monitoring systems and nearreal time applications) and accuracy, further demand the development of more sophisticated analysis concepts (Jain et al. 2000). Thus, the development of adequate methods for various data sets is an important ongoing research topic in the field of remote sensing. This chapter is organized as follows. In Section 1.1.2 a general introduction to machine learning is given, followed by a discussion on different paradigms in remote sensing. In Section 1.2 various supervised classifiers are introduced. A conclusion is given in Section 1.3. General concepts of machine learningMachine Learning is an area of artificial intelligence and generally refers to the development of methods that optimize their performance iteratively by learning from the data. Such methods can be predictive (e.g., a regression model) and make a prediction of a specific phenomenon or descriptive (e.g., a classification model) and distinguish for example between different classes of patterns. In the field of remote sensing, descriptive machine learning algorithms often focus on land cover classifications, and thus provide important input information in several environmental monitoring systems, as for instance in the area of flood forecast, urban sprawl and land degradation. In this cont...

show abstract

<title>Curvilinear component analysis for nonlinear dimensionality reduction of hyperspectral images</title>

Cited by 31 publications

References 0 publications

Hyperspectral Subspace Identification

Hyperspectral Subspace Identification

Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches

Machine Learning Techniques in Remote Sensing Data Analysis

Contact Info

Product

Resources

About