Dominique Barba

Compression of Polarimetric Synthetic Aperture Radar Data

The paper deals with proposition and evaluation of new and specific methods to represent vector radar data acquired by means a side-looking measurement in order to use compression process of Lind, Buzo, Gray (LBG), and Kohonen’s self organizing feature maps of topology. The aim is to enable after coding, transmission, and decoding a high-resolution reconstruction image using the Synthetic Aperture Radar (SAR) methods. The approach proposed for compression uses the statistical properties of the signals to be compressed in order to perform the vector quantification in an optimal way.

Comparative study of some algorithms for terrain classification using SAR images

This paper describes the use of polarimetric SAR data for Earth terrain identification and classification. The K-distribution is used to describe the spatial characteristics of a forest area. Statistical methods using supervised and unsupervised techniques are also applied to segment the simple and the polarimetric data. The supervised classification is based on the ML algorithm assuming a Rayleigh distribution for the intensity pixels of the HH image and a circular Gaussian for the full polarimetric data. For the unsupervised classification, we use both the clustering and the ICM (iterative conditional modes) algorithms. Using such techniques, the segmentation we could classify areas of trees which differ by their age and which have in general an approachable scattering behaviour.<<ETX>>

Vector quantization of raw polarimetric SAR data by using their statistical properties

Satellite radar imaging requires the transmission of a huge amount of data, since the image reconstruction can not usually be directly accomplished on board. A compression scheme is therefore needed. The current image resolutions do not allow lossless type of coding. The statistical properties of the signal are besides often discarded. However, the correlation of this signal can be exploited in order to increase the compression algorithms performances. This paper shows compression techniques examples and the contribution of the statistical properties of the signal for compression purpose.

Compression of Polarimetric Synthetic Aperture Radar Data — Abstract

The paper deals with proposition and evaluation of new and specific methods to represent vector radar data acquired by means a side-looking measurement in order to use compression process of Lind, Buzo, Gray (LBG), and Kohonens self organizing feature maps of topology. The aim is to enable after coding, transmission, and decoding a high-resolution reconstruction image using the Synthetic Aperture Radar (SAR) methods. The approach proposed for compression uses the statistical properties of the signals to be compressed in order to perform the vector quantification in an optimal way.

Filtrage du speckle dans les images ros par modification de contraste, comparaison avec une grande classe de filtres

Speckle appearing in synthetic aperture radar (sar) images is generated by coherent processing of radar signals. Basically the speckle has the nature of multiplicative noise. Recently, several methods have been proposed to remove speckle in sar images, among them, Lee’s methods, Frost’s method and Zamperoni’s method. In this paper, we propose an effective method for smoothing speckle-corrupted images. Digital processing is introduced, based on a contrast modification in a multi-resolution pyramid image representation. The contrast modification is realized by a median filter or an adaptive order filter giving a reconstructed image. For assessing this new method, we realized a comparative study with several noise-smoothing algorithms.RésuméLe phénomène du speckle est caractéristique des images laser, des images radiographiques et des images radar à ouverture synthétique (ros). Dans ces images le speckle se présente comme une granularité ayant les caractéristiques d’un bruit aléatoire multiplicatif. Plusieurs méthodes ont été proposées récemment pour la réduction du speckle dans les images ros, les principales étant celles de Lee, de Frost, de Zamperoni. Dans cet article, une nouvelle méthode de réduction du speckle est proposée. Celle-ci est basée sur la modification du contraste de l’image à partir d’une représentation multi-résolution pyramidale. La modification du contraste à chaque étage de la pyramide est réalisée à l’aide d’un filtre médian ou d’un filtre d’ordre adaptatif. L’image est reconstruite à partir du contraste modifié et engendre ainsi l’image filtrée. Pour évaluer les performances de cette nouvelle méthode, une étude comparative avec les principales méthodes existantes a été réalisée.Le phenomene du speckle est caracteristique des images laser, des images radiographiques et des images radar a ouverture synthetique (ros). Dans ces images le speckle se presente comme une granularite ayant les caracteristiques d’un bruit aleatoire multiplicatif. Plusieurs methodes ont ete proposees recemment pour la reduction du speckle dans les images ros, les principales etant celles de Lee, de Frost, de Zamperoni. Dans cet article, une nouvelle methode de reduction du speckle est proposee. Celle-ci est basee sur la modification du contraste de l’image a partir d’une representation multi-resolution pyramidale. La modification du contraste a chaque etage de la pyramide est realisee a l’aide d’un filtre median ou d’un filtre d’ordre adaptatif. L’image est reconstruite a partir du contraste modifie et engendre ainsi l’image filtree. Pour evaluer les performances de cette nouvelle methode, une etude comparative avec les principales methodes existantes a ete realisee.

Unsupervised optimal fuzzy clustering and Markov segmentation of polarimetric imaging

This paper presents a method for unsupervised segmentation of polarimetric SAR data into classes of homogeneous microwave backscatter characteristics. Clustering of polarimetric backscatter are obtained either by the CMF-NSO or be SEM algorithm. These algorithms carry out the classification without a priori assumptions on the number of classes in the data set. Assessment of cluster validity is based on performance measures using hypervolume V or CS function criteria. The later measures the overall average compactness and separation of a fuzzy-partition. The CMF-NSO algorithm performs well in situations of large variability of cluster shapes and densities. Given the clusters of polarimetric backscatter, the entire image is segmented using a MAP estimation. Implementation of the MAP technique is accomplished by an ICM algorithm. Results, using fully polarimetric SAR forest data, obtained by the CMF-NSO following by the ICM algorithm with a K-distribution model are quite satisfactory.

Adaptive vectorial speckle filtering in SAR images based on fuzzy clustering criteria

This paper deals with adaptive vector filtering of speckling in SAR images. The proposed method is based on the ordering of the vector data. The vector order is obtained by an Euclidean distance calculated from the center of a set of vectors. The local variation coefficient has been introduced for filter adaptivity. This coefficient is a reflection of the generalization of local scalar variation towards a case of a vectorial variation. For this purpose we use the fuzzy cluster analysis domain.In each large window, we decompose data into two groups using the fuzzy center mobile algorithm. The Euclidian distance between the two group centers provides good information about local variation in the window. In a homogeneous area this distance is minimal whereas it is considerable in the heterogeneous area. The global distribution of this distance coming from all windows of the image determine the threshold value. According to the local variation value, the vector will be treated in two different ways. If the local variation is less than the threshold value then we use a classic mean filtering. Otherwise, we replace the vector of interest by another ordered vector. Its choice depends on the local variation value.

Statistical information analysis of raw synthetic aperture radar data for compression

SAR data is a remote sensor which produces a high amount of data. Since the transmission of the corresponding raw signals requires extremely high data rates, on-board data compression is usually necessary. The approach presented in this paper use the statistic properties of the received signal in order to improve the compression performance.