Learning compatibility coefficients for relaxation labelingprocesses

Relaxation labeling processes have been widely used in many different domains including image processing, pattern recognition, and artificial intelligence. They are iterative procedures that aim at reducing local ambiguities and achieving global consistency through a parallel exploitation of contextual information, which is quantitatively expressed in terms of a set of “compatibility coefficients.” The problem of determining compatibility coefficients has received a considerable attention in the past and many heuristic, statistical-based methods have been suggested. In this paper, the authors propose a rather different viewpoint to solve this problem: they derive them attempting to optimize the performance of the relaxation algorithm over a sample of training data; no statistical interpretation is given: compatibility coefficients are simply interpreted as real numbers, for which performance is optimal. Experimental results over a novel application of relaxation are given, which prove the effectiveness of the proposed approach     

MST-based stepwise connection strategies for multipass Radar data, with application to coregistration and equalization

This paper proposes a unified framework for predicting optimized pairing strategies for interferometric processing of multipass synthetic aperture radar data. The approach consists in a minimum spanning tree (MST) structure based on a distance function encoding an a priori model for the interferometric quality of each image pair. Using a distance function modeled after the interferometric coherence allows reproducing many "small baseline" strategies presented in the recent literature. A novel application of the method to the processing steps of image coregistration and equalization is illustrated, using a test European Remote Sensing Satellite dataset. Widespread methods used for these two operations rely on the computation of the amplitude cross correlation over a large number of corresponding tie patches distributed over the scene. Geometric shift and radiometric equalization parameters are estimated over the patches and used, respectively, within a polynomial warp model and a radiometric correction scheme. The number of reliable patches available behaves similarly to the interferometric synthetic aperture radar (InSAR) coherence with respect to the baselines, and can be assimilated to a quality figure for the derivation of the MST. Results show an improvement in the quality of the stepwise (SW)-processed image stack with respect to the classical single-master procedure, confirming that the SW approach is able to provide better conditions for the estimation of correlation-related InSAR parameters.     

A Wide-Band Approach to the Absolute Phase Retrieval in SAR Interferometry

Because of possible multiple solutions allowed, the unwrapping of interferometric fringe patterns in the spatial domain is an ill-posed problem which needs some a priori knowledge of the ground morphology for the solution of ambiguities. This is especially true for interferometric SAR (Synthetic Aperture Radar) data. In this paper we propose a different approach to InSAR processing for retrieving the height of ground points independently from each other, unlike most conventional phase unwrapping procedures, which operate in the spatial domain. The basic idea is to repeat raw data focusing by using range sub-bands centered at different frequencies, in order to find a point history of the interferometric phase variation vs. frequency. We introduce the general framework of the method together with considerations on the theoretical limits of applicability, then we report results of our simulations related to a wide-band SAR system. We show that, under certain conditions, height values can be retrieved over a network of coherent and strong scatterers, even when enclosed into low-coherence areas.     

Comparison of SAR amplitude vs. coherence flood detection methods - a GIS application

Flood area detection from multipass Synthetic Aperture Radar (SAR) data can be performed via amplitude change detection techniques. These methods allow flooded zones to be discriminated only when they are flooded at the time of the second passage, and not at the time of the first one. Coherence derived from multipass SAR interferometry can be used instead, as an indicator of changes in the electromagnetic scattering behaviour of the surface, thus potentially revealing all the areas affected by the flood event at any time between the two passes. The paper presents a prototype application of such techniques, that is, a flood map obtained from ERS-1/2 data taken over Beziers (Southern France), through proper thresholding of a combination of amplitude and coherence information. Produced in the framework of an ESA project, the map consists of a DXF vector file which can be imported directly into most commercial GIS software.     

Unwrapping-free interpolation of sparse DInSAR phase data: experimental validation

In multi-temporal applications of synthetic aperture radar (SAR) interferometry, differential phase contributions due to atmospheric inhomogeneities, estimated over sparse points, have to be interpolated and removed from the regular-grid interferograms in order to highlight the phase stability of more image pixels, which then add to the available data to infer useful information about terrain displacements or other phenomena of interest. Interpolation is usually done on the phase data after a phase unwrapping (PU) operation. In a previous work, we considered the alternative interpolation step applied directly to the complex phasor derived from the wrapped phase, thus bypassing the error-prone sparse PU operation. In this article, the performances of the proposed methodology are evaluated over atmospheric phase screen (APS) data estimated from a previous processing through persistent scatterers interferometry (PSI) methods. The original persistent scatterer (PS) population is reduced by thresholding their inter-image coherence values, and then further subsampled randomly in a rectangle inside a detected subsidence bowl. Both the classical and the proposed interpolation procedures are applied to the subsampled APS phase values. The interpolated fields are then removed from the rest of the PS, and the residual phase values are compared in terms of inter-image coherence. Results confirm that interpolating complex phasors, thus avoiding PU, gives results equivalent to the standard procedure in good sampling conditions. Moreover, when point sparsity induces phase aliasing, thus hindering the PU operation, the proposed method allows to better recover phase information over unsampled pixels, improving the final results of the PSI processing.