Search

In this work, unsupervised change detection techniques, based on three different way to compare images, are presented. Two Landsat TM registered and corrected multi-spectral images, acquired on the same geographical area on 18 May 1996 and 21 May 1997, have been used. In the first comparison technique, for each pair of corresponding pixels, the spectral change vector has been computed as the squared difference in the features vectors at the two times. In the second method, the difference image has been computed using, pixel by pixel, a chi square transformation. The third technique is based on the application of a Self-Organizing Map (SOM) neural network to clusterize the two images before comparison. The three obtained difference images has been then analyzed by using a fully automatic thresholding method exploiting the expectation-maximization (EM) algorithm. The experimental results obtained for the three difference images are comparable, showing a reliable robustness of the unsupervised approach, and only few change are detected on the analyzed scene. Moreover, the experimental results have been compared with a change detection map computed by using a supervised technique, obtaining a good agreement between unsupervised and supervised results that confirms the reliability of the considered approach. The encouraging obtained results allow to use the so-computed percentage value of changes as probability of class transitions in input to a Bayesian supervised change detection method, as presented in a companion paper by the same authors. In this framework, the unsupervised approach may be used to support supervised techniques, providing land cover transitions that can be used as guess values

The purpose of this paper is to investigate the possibility of estimating soil moisture content from single angle and single frequency SAR data. To reach this objective, a priori information such as profile height std is assumed. An inversion technique based on a neural approach trained with the IEM model is presented and its performances are discussed. Two sources of errors in the final estimates are analyzed. Errors due to the limited input information and errors due to possible failures of IEM in predicting scattering from natural surfaces.

High resolution extended DEM can be obtained from interferometric SAR image pairs. In order to evaluate the elevation of a pixel with respect to a slant-range reference plane, its absolute interferometric phase is required and this is usually derived unwrapping a 2D interferometric fringe pattern. Different methods have been suggested for phase unwrapping, based on local or global approaches. In both cases, phase inconsistencies (residues) in SAR interferograms due to noise and topography determine unwrapping errors which appear as discontinuities in the computed absolute phase field. In this paper, results are reported on the authors' experiment addressed to improve the output of phase unwrapping techniques by means of a residue classification procedure, according to their origin. Working on both simulated and real phase fields, the authors have found some differences between the spatial configurations of phase fields around singular points due to noise or linear discontinuities. A practical method has been developed to solve some ambiguities in the phase unwrapping of real SAR interferograms, useful for the linear integration approach as well as for global methods.

The work presents preliminary results obtained in a project aimed to analyse the possibility to use different approaches in the analysis of multisensor and multitemporal data for geologic applications. The study area is characterised by alluvial and marine deposits, mainly consisting of clays, sandstones and conglomerates, and is characterised by slowly progressing landslides. In order to evaluate the complementarity of optical and microwave data for geological inspection, an analysis with Landsat Thematic Mapper (TM) data is performed by visual inspection. Afterwards the ERS-1 (Earth Remote Sensing Satellite) multitemporal data set is considered in the same way. The results consist of lineament maps that present different structures, demonstrating the complementarity of the two data set. Lineament maps are digitized and statistically compared, in term of frequency of direction. >

This paper presents a comparative evaluation between two classification strategies for the analysis of remote sensed data. The first is based on the combination of the outputs of a neural network (NN) ensemble, the second concerns the application of Support Vector Machine (SVM) classifiers. Emphasis is given to the understanding of the limits and the advantages of the two strategies to design a classifier system able to provide high generalization capability. Two sets of experiments have been carried out to classify benchmark remote sensed data. In the first set a Fuzzy Integral has been used to combine the outputs of neural classifiers in an ensemble. In the second set of experiments SVM classifiers have been trained and tested on the same data set. The comparative analysis evidences that SVM classifiers outperform an ensemble of classifiers, whose partial results are combined by a Fuzzy Integral. The training complexity of SVMs seems, however, to be a limitation to the extensive use of SVMs in complex multisource-multitemporal data classification. © Springer-Verlag Berlin Heidelberg 2006.

Microwave sensors are of primary importance in mapping surface states and measuring some significant quantities which affect the hydrological cycle. A space mission aiming at monitoring soil moisture and surface salinity at a global scale is suggested. The mission is based on a combination of polarimetric active and passive microwave sensors.

The detection of marine oil slicks using satellite sun-glittered optical imagery has been recently assessed. As the nature of the imaging mechanism involves the altered features of the wind-roughened oil-covered sea surface, it is expected that the radiation reflected from the oil-water system carries information about the physical properties of the floating oil layer. In this paper, we report an investigation on the capability to retrieve the average thickness of thin marine oil slicks by using the sun-glittered component of the solar radiation in the near-infrared (NIR) bands of MEdium Resolution Imaging Spectrometer Instrument (MERIS) and MODerate Resolution Imaging Spectroradiometer (MODIS) images. The developed procedure exploits the Cox and Munk model to compute sun glint reflectance at the sea surface level for both clean and oil polluted sea surface as well. It is assumed that the Fresnel reflection coefficient of the oil-water system carries the relevant optical dependence on oil layer thickness and oil type. The expected oil-water system reflectance is computed by taking into account the non-uniform spatial distribution of the oil volume. This is achieved by considering a pdf of oil thicknesses that matches the observations on controlled oil slicks already reported in the scientific literature. MERIS and MODIS images gathered during the Lebanon oil spill occurred on July and August 2006 were selected as case study. When available, co-located SAR imagery was also considered to corroborate NIR-detected oil slicks.