This paper examines the utility of hyperspectral imagery for remote sensing data analysis. The acquired data volumes are very important, often reaching hundreds or thousands of channels for a single scene observed. Certainly, the large quantity contained in the hyperspectral database is accompanied by a complex physical content and consequently a considerable time computing which can affect the quality of treatment. These channels come from a very fine spectral sampling, allow discriminating and differentiating constituents that are spectrally close. Furthermore, hyperspectral imaging is a technique that is as strong potential initiating new research in the development of mathematical morphology. This theory, mainly inspired by the image processing problems, extends to a new more complex scope of which is hyperspectral mathematical morphology. Applied to hyperspectral data in hyper dimensional features spaces, we compare two proposed classification approaches. The first method is based on centralized segmentation methodology which exploits the information complementarities. The second method is based on hierarchical clustering which consists on combining a divisive clustering approach applied at a high-level and an agglomerative clustering operating at a low-level.

High-dimensional data applications have earned great attention in recent years. We focus on remote sensing data analysis on high-dimensional space like hyperspectral data. From a methodological viewpoint, remote sensing data analysis is not a trivial task. Its complexity is caused by many factors, such as large spectral or spatial variability as well as the curse of dimensionality. The latter describes the problem of data sparseness. In this particular ill-posed problem, a reliable classification approach requires appropriate modeling of the classification process. The proposed approach is based on a hierarchical clustering algorithm in order to deal with remote sensing data in high-dimensional space. Indeed, one obvious method to perform dimensionality reduction is to use the independent component analysis process as a preprocessing step. The first particularity of our method is the special structure of its cluster tree. Most of the hierarchical algorithms associate leaves to individual clusters, and start from a large number of individual classes equal to the number of pixels; however, in our approach, leaves are associated with the most relevant sources which are represented according to mutually independent axes to specifically represent some land covers associated with a limited number of clusters. These sources contribute to the refinement of the clustering by providing complementary rather than redundant information. The second particularity of our approach is that at each level of the cluster tree, we combine both a high-level divisive clustering and a low-level agglomerative clustering. This approach reduces the computational cost since the high-level divisive clustering is controlled by a simple Boolean operator, and optimizes the clustering results since the low-level agglomerative clustering is guided by the most relevant independent sources. Then at each new step we obtain a new finer partition that will participate in the clustering process to enhance semantic capabilities and give good identification rates. (C) 2016 Society of Photo-Optical Instrumentation Engineers (SPIE)

In this paper we address the problem of the remote sensing data analysis on high-dimensional space like hyperspectral data. Its complexity is caused by many factors, such as the large spectral or spatial variability and the curse of dimensionality. Much work has been carried out in the literature to overcome this particular ill-posed issue. Applied to hyperspectral data in hyperdimensional features spaces, the first particularity of our method is the special structure of its cluster tree. Their leaves are associated with sources represented according to mutually independent axes to represent specifically some land covers. The second particularity is that at each level of the cluster tree we combine a divisive clustering approach applied at a high-level and an agglomerative clustering operating at a low-level. We propose to compare the performances of our approach with those of hierarchical method based on Ward approach.

In this paper, we consider the problem of blind image separation by taking advantage of the sparse representation of the hyperspectral images in the DCT-domain. Blind Source Separation (BSS) is an important field of research in signal and image processing. These images are produced by sensors which provide hundreds of narrow and adjacent spectral bands. The idea behind transform domain is that we can restructure the signal/image values to give transform coefficients more easily to separate. This work describes a novel approach based on Second-Order Separation by Frequency-Decomposition, termed SOSFD. This technique uses joint information from second-order statistics and sparseness decomposition. Furthermore, the proposed approach has the added advantages of the DCT and second-order statistics in order to select the optimum data information. In fact, representing the hyperspectral images in well suited database functions allows a good distinction of various types of objects. Results show the contribution of this new approach for the hyperspectral image analysis and prove the performance of the SOSFD algorithm for hyperspectral image classification. On the opposite of the original images that are represented according to correlated axes, the source images extracted from the proposed approach are represented according to mutually independent axes that allow a more efficient representation of information contained in each image. Then, each source can represent specifically certain themes by exploiting the link between the frequency-distribution and structural composition of the image. This application is of utmost importance in the classification process and could increase the reliability of the analysis and the interpretation of the hyperspectral images.

In this paper, we consider the problem of Blind source separation (BSS) method by taking advantage of the sparse modeling of the hyperspectral images. These images are produced by sensors which provide hundreds of narrow and adjacent spectral bands. The idea behind transform domains is to apply some transformations to illustrate the dataset with a minimum of components and a maximum of essential information. To take advantages from the new representation of hyperspectral data, a novel classification approach based on using Binary Partition Trees (BPT). The BPT is obtained by iteratively merging regions and provided a combined and hierarchical representation of the image in a tree structure of regions.

Source separation is relatively a new area of data analysis. The most widely used separation approach's are linear. However, in many realistic cases the process which generates the observations is nonlinear and no information is available about the mixture. In this case, it can be expected to capture the structure of the data better if the data points lie in a nonlinear manifold instead of a linear subspace. In this paper, we try to find a model which allows a compact description of the observations in the hope of discovering some of the underlying causes or sources of the observations. Then, we will process a dimension reduction to classify the obtained sources and evaluate the performances of the proposed method.

In this paper, we consider the problem of blind image separation by taking advantage of the sparse representation of the study images in the DCT-domain. Blind source separation (BSS) is an important field of research in signal and image processing. The BSS problem has been considered either directly in the original domain of observations or in a transform domain. The idea behind transform domains is that usually an invertible linear transform restructures the signal/image values to give transform coefficients more easily to separate. This paper describes a new method for blind source separation. The latter takes advantage of the sparse representation of structured data in large overcomplete dictionaries to separate independent features. Furthermore, DCT exhibits excellent energy compaction for highly correlated images such as hyperspectral images, which permits to reduce significantly the complexity of the separation. For this purpose, we will exploit the redundancy of neighboring pixels and the correlation of adjacent bands by a new source separation approach based jointly on the Blind Source Separation (BSS) and Discrete Cosine Transform (DCT). In this work, we differentiate from the previous works by using a second order source separation criterion in the frequency domain. The extracted independent components may lead to a meaningful data representation which permits to extract information at a finer level of precision. This approach is of utmost importance in the classification process and should minimize the misclassification risk of hyperspectral images.