Congreso: XXIII Simposium Nacional de la Unión Científica Internacional de Radio (URSI 2008). 22-24 septiembre 2008. Madrid, España.

Autores: José Tomás González Partida, Mateo Burgos García, Mihai Datcu

This article presents a method for the determinationof the relative misregistration between two interferometric syntheticaperture radar (InSAR) images. Classical coregistrationtechniques are based on the maximization of the crosscorrelationcoefficient, i.e. the maximum likelihood (ML)estimate of the coherence magnitude. This paper proposes anefficient algorithm to solve the coherence maximization withhigh accuracy, reducing the computational complexity of theclassical techniques based on oversampled data. Furthermore,several coherence estimators have been proposed in theliterature in addition to the ML estimate. This paper evaluatesthe behaviour of the algorithm using three different coherencemagnitude estimators as the kernel of the optimization problem.The discrete Fourier transform (DFT) is exploited to formulatethe algorithm and to explain the necessity to register the imageswith subpixel accuracy. The performance of the technique hasbeen proved over simulated data of a high resolutionmillimeter-wave InSAR system that is currently underdevelopment.