Fusion of LIDAR and POLINSAR images for forest vertical structure retrieval

Abstract:
This paper presents the overall strategy of fusion of LIDAR images and P-band PolInSAR images of forests in order to exploit their synergistic information. The images considered are from the 2010 BioSAR campaign on Remningstorp forest in Sweden. As a first step, we propose a pixel to pixel coregistration method of the images, and we assess the accuracy of this coregistration. Then, we propose descriptors for a given pixel, for each type of images. The primary challenge is to find features that go beyond the difference of geometrical configurations between the two types of information, and also that compensate the effect of the incidence angle on radar observables. As of today, we propose to analyse the combination of PolInSAR and lidar feature in order to see their variations depending on radar acquisition conditions and scatterer properties. We also test a support vector regression on two selected radar and lidar feature in order to draw some conclusions and improvements points.