Regularizationof Displacement Fields for Glaciermonitoring

Synthetic aperture radar (SAR) images provide scattering information which can be used under any weather conditions for glacier monitoring. The purpose is to estimate a displacement field characterizing at each position the local speeds and orientations of the glacier movement. The recent method proposed in [1] builds a vector field from the similarities existing between two SAR images co-registered on static areas and sensed at different time. The local movement is evaluated by researching in one acquisition the patterns which are similar to those present in the other acquisition. This abstract proposes to use the correspondences between patterns inside a two step estimator of the vector field. The first step is the local estimation of the displacement vectors using the maximum likelihood estimator. The presented results show that this local estimator seems to be efficient in textured area. However, the spatial coherence of the vector field is not insured in homogeneous area. Then, a second step of regularization seems to be necessary to deal with such unreliable estimated vectors. Different regularization methods, which could be used in the final paper, are proposed to improve the spatial coherence of the displacement field.