Nonlinear mixture model for hyperspectral unmixing

Proceedings of International Conference - SPIE 7477, Image and Signal Processing for Remote Sensing XV - 28 September 2009

Performance analysis using equivariant kernel density estimator in nonlinear mixture

This paper investigates the performance analysis of separation of mutually independent sources in nonlinear models. The nonlinear mapping constituted by an unsupervised linear mixture is followed by an unknown and invertible nonlinear distortion, are found in many signal processing cases. Generally, blind separation of sources from their nonlinear mixtures is rather difficult. We propose using a kernel density estimator incorporated with equivariant gradient analysis to separate the sources with nonlinear distortion. The kernel density estimator parameters of which are iteratively updated to minimize the output independence expressed as a mutual information criterion. The equivariant gradient algorithm has the form of nonlinear decorrelation to perform the convergence analysis. Experiments are proposed to illustrate these results.

Monitoring the composition and form of urban environments based on the vegetation-impervious surface-soil (VIS) model by sub-pixel analysis techniques

The majority of the world's population now resides in urban environments and information on the internal composition and dynamics of these environments is essential to enable preservation of certain standards of living. Remotely sensed data, especially the global coverage of moderate spatial resolution satellites such as Landsat, Indian Resource Satellite and Systeme Pour I'Observation de la Terre (SPOT), offer a highly useful data source for mapping the composition of these cities and examining their changes over time. The utility and range of applications for remotely sensed data in urban environments could be improved with a more appropriate conceptual model relating urban environments to the sampling resolutions of imaging sensors and processing routines. Hence, the aim of this work was to take the Vegetation-Impervious surface-Soil (VIS) model of urban composition and match it with the most appropriate image processing methodology to deliver information on VIS composition for urban environments. Several approaches were evaluated for mapping the urban composition of Brisbane city (south-cast Queensland, Australia) using Landsat 5 Thematic Mapper data and 1:5000 aerial photographs. The methods evaluated were: image classification; interpretation of aerial photographs; and constrained linear mixture analysis. Over 900 reference sample points on four transects were extracted from the aerial photographs and used as a basis to check output of the classification and mixture analysis. Distinctive zonations of VIS related to urban composition were found in the per-pixel classification and aggregated air-photo interpretation; however, significant spectral confusion also resulted between classes. In contrast, the VIS fraction images produced from the mixture analysis enabled distinctive densities of commercial, industrial and residential zones within the city to be clearly defined, based on their relative amount of vegetation cover. The soil fraction image served as an index for areas being (re)developed. The logical match of a low (L)-resolution, spectral mixture analysis approach with the moderate spatial resolution image data, ensured the processing model matched the spectrally heterogeneous nature of the urban environments at the scale of Landsat Thematic Mapper data.

Learning dependent sources using mixtures of Dirichlet: applications on hyperspectral unmixing

This paper is an elaboration of the DECA algorithm [1] to blindly unmix hyperspectral data. The underlying mixing model is linear, meaning that each pixel is a linear mixture of the endmembers signatures weighted by the correspondent abundance fractions. The proposed method, as DECA, is tailored to highly mixed mixtures in which the geometric based approaches fail to identify the simplex of minimum volume enclosing the observed spectral vectors. We resort then to a statitistical framework, where the abundance fractions are modeled as mixtures of Dirichlet densities, thus enforcing the constraints on abundance fractions imposed by the acquisition process, namely non-negativity and constant sum. With respect to DECA, we introduce two improvements: 1) the number of Dirichlet modes are inferred based on the minimum description length (MDL) principle; 2) The generalized expectation maximization (GEM) algorithm we adopt to infer the model parameters is improved by using alternating minimization and augmented Lagrangian methods to compute the mixing matrix. The effectiveness of the proposed algorithm is illustrated with simulated and read data.

Hyperspectral subspace identification

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.

Does independent component analysis play a role in unmixing hyperspectral data?

Independent component analysis (ICA) has recently been proposed as a tool to unmix hyperspectral data. ICA is founded on two assumptions: 1) the observed spectrum vector is a linear mixture of the constituent spectra (endmember spectra) weighted by the correspondent abundance fractions (sources); 2)sources are statistically independent. Independent factor analysis (IFA) extends ICA to linear mixtures of independent sources immersed in noise. Concerning hyperspectral data, the first assumption is valid whenever the multiple scattering among the distinct constituent substances (endmembers) is negligible, and the surface is partitioned according to the fractional abundances. The second assumption, however, is violated, since the sum of abundance fractions associated to each pixel is constant due to physical constraints in the data acquisition process. Thus, sources cannot be statistically independent, this compromising the performance of ICA/IFA algorithms in hyperspectral unmixing. This paper studies the impact of hyperspectral source statistical dependence on ICA and IFA performances. We conclude that the accuracy of these methods tends to improve with the increase of the signature variability, of the number of endmembers, and of the signal-to-noise ratio. In any case, there are always endmembers incorrectly unmixed. We arrive to this conclusion by minimizing the mutual information of simulated and real hyperspectral mixtures. The computation of mutual information is based on fitting mixtures of Gaussians to the observed data. A method to sort ICA and IFA estimates in terms of the likelihood of being correctly unmixed is proposed.

Does Independent Component Analysis Play a Role in Unmixing Hyperspectral Data?

Chapter in Book Proceedings with Peer Review First Iberian Conference, IbPRIA 2003, Puerto de Andratx, Mallorca, Spain, JUne 4-6, 2003. Proceedings

Caracterização da ocupação do solo com recurso à aplicação de modelos de misturas espectrais em séries multi-temporais de imagens MODIS

Dissertação apresentada como requisito parcial para obtenção do grau de Mestre em Ciência e Sistemas de Informação Geográfica

Estimation of signal subspace on hyperspectral data

Dimensionality reduction plays a crucial role in many hyperspectral data processing and analysis algorithms. This paper proposes a new mean squared error based approach to determine the signal subspace in hyperspectral imagery. The method first estimates the signal and noise correlations matrices, then it selects the subset of eigenvalues that best represents the signal subspace in the least square sense. The effectiveness of the proposed method is illustrated using simulated and real hyperspectral images.

Parallel method for sparse semisupervised hyperspectral unmixing

Parallel hyperspectral unmixing problem is considered in this paper. A semisupervised approach is developed under the linear mixture model, where the abundance's physical constraints are taken into account. The proposed approach relies on the increasing availability of spectral libraries of materials measured on the ground instead of resorting to endmember extraction methods. Since Libraries are potentially very large and hyperspectral datasets are of high dimensionality a parallel implementation in a pixel-by-pixel fashion is derived to properly exploits the graphics processing units (GPU) architecture at low level, thus taking full advantage of the computational power of GPUs. Experimental results obtained for real hyperspectral datasets reveal significant speedup factors, up to 164 times, with regards to optimized serial implementation.

Blind hyperspectral unmixing

Hyperspectral unmixing methods aim at the decomposition of a hyperspectral image into a collection endmember signatures, i.e., the radiance or reflectance of the materials present in the scene, and the correspondent abundance fractions at each pixel in the image. This paper introduces a new unmixing method termed dependent component analysis (DECA). This method is blind and fully automatic and it overcomes the limitations of unmixing methods based on Independent Component Analysis (ICA) and on geometrical based approaches. DECA is based on the linear mixture model, i.e., each pixel is a linear mixture of the endmembers signatures weighted by the correspondent abundance fractions. These abundances are modeled as mixtures of Dirichlet densities, thus enforcing the non-negativity and constant sum constraints, imposed by the acquisition process. The endmembers signatures are inferred by a generalized expectation-maximization (GEM) type algorithm. The paper illustrates the effectiveness of DECA on synthetic and real hyperspectral images.

Hyperspectral unmixing algorithm via dependent component analysis

This paper introduces a new method to blindly unmix hyperspectral data, termed dependent component analysis (DECA). This method decomposes a hyperspectral images into a collection of reflectance (or radiance) spectra of the materials present in the scene (endmember signatures) and the corresponding abundance fractions at each pixel. DECA assumes that each pixel is a linear mixture of the endmembers signatures weighted by the correspondent abundance fractions. These abudances are modeled as mixtures of Dirichlet densities, thus enforcing the constraints on abundance fractions imposed by the acquisition process, namely non-negativity and constant sum. The mixing matrix is inferred by a generalized expectation-maximization (GEM) type algorithm. This method overcomes the limitations of unmixing methods based on Independent Component Analysis (ICA) and on geometrical based approaches. The effectiveness of the proposed method is illustrated using simulated data based on U.S.G.S. laboratory spectra and real hyperspectral data collected by the AVIRIS sensor over Cuprite, Nevada.

The influence of the conservation status and changes in the rainfall regime on forest-savanna mosaic dynamics in Northern Brazilian Amazonia

Coupled carbon/climate models are predicting changes in Amazon carbon and water cycles for the near future, with conversion of forest into savanna-like vegetation. However, empirical data to support these models are still scarce for Amazon. Facing this scenario, we investigated whether conservation status and changes in rainfall regime have influenced the forest-savanna mosaic over 20 years, from 1986 to 2006, in a transitional area in Northern Amazonia. By applying a spectral linear mixture model to a Landsat-5-TM time series, we identified protected savanna enclaves within a strictly protected nature reserve (Maracá Ecological Station - MES) and non-protected forest islands at its outskirts and compared their areas among 1986/1994/2006. The protected savanna enclaves decreased 26% in the 20-years period at an average rate of 0.131 ha year-1, with a greater reduction rate observed during times of higher precipitation, whereas the non-protected forest islands remained stable throughout the period of study, balancing the encroachment of forests into the savanna during humid periods and savannization during reduced rainfall periods. Thus, keeping favorable climate conditions, the MES conservation status would continue to favor the forest encroachment upon savanna, while the non-protected outskirt areas would remain resilient to disturbance regimes. However, if the increases in the frequency of dry periods predicted by climate models for this region are confirmed, future changes in extension and directions of forest limits will be affected, disrupting ecological services as carbon storage and the maintenance of local biodiversity.