Digital spectral analysis: with applications
Digital spectral analysis: with applications
| Hi-index | 0.02 |
This paper deals with the application of modern spectral analysis techniques to polarimetric Synthetic Aperture Radar (SAR) data. The purpose (Section 1) is to use all the information provided by images of the same SAR scene at the different polarimetric channels (HH, VV, HV and VH) in order to extract all the information on the scene back-scattering properties with better geometrical resolution than the resolution corresponding to the compressed coded waveform and the synthetic aperture. Super-resolved images are obtained by replacing the classical spectral estimator (i.e. the Fast Fourier Transform, FFT) with parametric spectral estimators such as those built around an autoregressive model of the dechirped signal. The proposed processing schemes--based on two-dimensional covariance method--are presented in Section 2 for both single channel data and polarimetric data. The joint use of different polarimetric images requires the choice of a suitable fusion technique to obtain a single super-resolved SAR image: to this aim both a decentralised fusion strategy (fusion of separately super-resolved images) and a centralised fusion strategy (super-resolution of the fused image) are proposed. The application of the techniques to polarimetric SIR-C SAR images of a ship is presented in Section 3 where the results obtained by using single channel images and multipolarisation images are discussed and compared. A perspective for future research and applications (Section 4) closes the paper.