Discretization Effects in Sea Surface Simulation Applied to Ship Classification Studies
Gerard Margarit(1) and Jordi J. Mallorqui(1)
(1) Universitat Politecnica de Catalunya (UPC), D3-Campus Nord-UPC, 08034, Spain
In the last years, ship classification has been significantly improved. Analysis of real and simulated data has allowed to better understand the reflectivity properties of vessels and how they can be processed for geometry discrimination. In this framework, the numerical tool GRECOSAR that combines an electromagnetic solver (GRECO) plus a SAR simulator has appeared to be very useful . In the past, GRECOSAR has been used to generate large databases of controlled SAR, PolSAR and PolInSAR images for different types of ships, imaging sensors and environmental conditions. With such images, new knowledge about ship scattering has been obtained from which a novel ship identification method based on PolInSAR imagery has been developed . First tests in simulated scenarios have shown reliable performance even against image distortions induced by ship motions and sea-ship interaction.
But these results are preliminary and the reliability of the method for ship classification has not been fully assessed. The main problem is the limited sea surface model of GRECOSAR that avoids to generate reliable clutter and reach real-like observation conditions. In this context, the current paper will try to overcome this drawback. For such purpose, sea surfaces modelled with facets and following the two-scale and sea spectrum approaches are considered. They are integrated in the programming structure of the simulator so that ship and sea responses can be computed at the same time. The paper focuses on evaluating which discretization requirements have to be fulfilled for achieving the best trade-off between clutter realism and processing time. Different models of open seas are evaluated for diverse facet lengths and operating frequencies so that the advisable ratio between facet length and operating wavelength may be properly estimated. The reliability of the resulting clutter is evaluated by comparing the Probability Density Function estimated via image histogram with typical distributions associated to real sea scenes. Among different options, Rayleigh PDF is used as a quality descriptor because it appears to reasonably describe the probabilistic properties of the sea .
The influence of different target discretization parameters in sea clutter performance is also inspected. This will allow to know which parameters demand more restrictive values and which one does not need special care. The analysis of all the presented data will show that it is possible to simulate reliable clutter useful to study the impact of the sea on phase stability and polarimetric mechanism, and, in turn, on the identification performance. In this framework, the novel identification method is tested under these new environmental conditions closer to the real ones. The identification ratios appear to be quite good increasing the potentiality of the method for its usage in real scenarios.
 G. Margarit and et al, “On the usage of GRECOSAR, an orbital polarimetric SAR simulator of complex targets, for vessel classification studies,” IEEE Trans. Geosci. Remote Sensing, vol. 44, no. 12, pp. 3517–3526, Dec. 2006.
 G. Margarit, J. J. Mallorqui, and X. Fabregas, “Single-pass polarimetric SAR interferometry for vessel classification,” IEEE Trans. Geosci. Remote Sensing, To be published in EUSAR Special Issue November 2007.
 J. V. Toporkov and M. A. Sletten, “Statistical Properties of Low-Grazing Range-Resolved Sea Surface Backscatter Generated Through Two-Dimensional Direct Numerical Simulations”, IEEE Trans. Geosci. Remote Sensing, vol. 45, no. 5, pp. 1181-1197, May 2007.
Keywords: ESA European
Space Agency - Agence spatiale europeenne,
observation de la terre, earth observation,
satellite remote sensing,
teledetection, geophysique, altimetrie, radar,
chimique atmospherique, geophysics, altimetry, radar,