Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal LAN0638

TitleThe use of TerraSAR-X quad and dual polarimetric data for target detection with a novel algorithm exploiting the polarization fork
Investigator Cloude, Shane - AEL Consultants, N/A
Team Member
Mr. Marino, Armando - The University of Edinburgh,, Edinburgh Earth Observatory (EEO),
Dr. Chen, Erxue - Chinese Academy of Forestry, Institute of Forest Resources Information techniques
Dr. Hong, Wen - Chinese Academy of Sciences, Institute of Electronics, (IECAS),
SummaryThe main objective of the project is the validation of a novel polarimetric detector (first presented in POLinSAR09 [1]) with satellite data. Considering the high resolution and different polarimetric modes, TerraSAR-X represents the best candidate for our polarimetric target detection. This is based on the potential to extract the target of interest in the target complex space with a filter, which uses a set of reference scattering mechanisms and generates a complex signal pair by perturbation of this reference set using geometrical rules derived from the Poincare’ sphere representation of target scattering. The coherence of this pair is then used to detect the desired feature, high coherence being associated with a good match and low coherence with a bad. The proposal can be basically divided in three main parts. Firstly, we will use the demonstration T-X quad pol data in order to use the detector in its classical quadpol way. The capability of the detector was already validated on L-band quad pol data, consequently reasonable results should be assured. In this project we want to undertake a quantitative assessment of the impact of satellite radar parameters, such as wavelength and resolution on filter performance at X-band. Secondly, the algorithm originally developed for quad pol data has been modified in order to perform the detection with dual pol data (the results are not yet published due to the innovativeness of the algorithm but in the following section we present some preliminary results). Consequently, dual pol HH/VV data could be used to validate the new technique as well as the possibility of using TerraSAR-X for satellite polarimetric detection (even if just dual pol data are required). The essential idea of the dual pol detection is that with two polarisation that can represent two orthogonal targets (i.e. vectors) of the polarimetric space a target subspace can be reconstructed, which contains the target of interest. For this task, the location of the scene constitutes a key point, since an area with several artificial targets (i.e. urban area) is to be preferred. Third, we want to develop applications of the detector for forestry purposes. The detector could be trained to investigate particular structures within the canopy or on the ground (as long as the forest is sparse enough). Moreover, the use of X band can be advantageous on the presence of sparse woodlands since the small wavelength allows examination of smaller plant structures that could be missed with longer wavelengths like L band. Consequently, we would like to apply for scenes where presence of forest/woodlands is quite consistent We would also make use of ground truth campaigns to be carried out during the project in order to validate the presence of target of interest in the scene. The campaigns are prepared in collaboration with the Chinese Academy of Forestry and Chinese Academy of Sciences through the Dragon program. The latter also plan to overfly the Chinese test site with an airborne fully polarimetric X-Band airborne sensor for comparisons with T-X data. In this context, the Taian test area is a unique opportunity, since it presents urban area beside forests where extensive ground truth campaigns were carried out. On the other side, Glen Affric is a very well studied test area where the presence of sparse woodland is quite abundant.

Back to list of proposals

© DLR 2004-2016