This work introduces a general method for glaciers surface displacement estimation using High Resolution PolSAR images. With high resolution details, the texture tracking methods become useful and complementary to the classical differential interferometry methods which are difficult to implement in this context. The proposed method is based on the multivariate SAR images ratio likelihood. The use of the ratio and corresponding statistic distributions is well adapted to the multiplicative noise of such images. The extracted texture component from HR PolSAR data according to SIRV model is less sensitive to the variations of the snow surface than the classical intensity. The statistical modeling of such component is discussed and the benefit of Fisher modeling is highlighted. Some tests have been performed to analyze the behavior of various similarity criteria. The methodology is finally completed with advanced processing like hierarchical segmentation and the use of a flow model. Two test sites in French Alps with the Argentière glacier and in Antarctica with the Astrolabe glacier have been used to validate the proposed method. A preliminary study on sensor merging for displacement estimation is finally introduced.
