Airborne scan mode ISAR imagery of ships using high resolution spectral methods and particle filter

The basic principle of Inverse Synthetic Aperture Radar (ISAR) imaging consists of performing a Doppler analysis of the signal returned by the target. In the context of maritime surveillance mission, the Doppler analysis, which is usually performed via FFT-based methods, enables to form a profile view of the ship. We propose to form the ISAR image while antenna is scanning via high resolution spectral estimation methods. The goal of using superresolution algorithms instead of Fourier methods is to improve resolution and to remove sidelobe, especially in our context where the rotation of the antenna implies a shorter dimension of the processed signals. We present in this article results of the application of such methods on both synthetic and real data. Next, this paper considers the possibility to track the scattering centers of the ISAR image along the different antenna scans. This technique will allow a better estimation of discriminating parameters in an additional classification step. We propose to use particle filter methods to perform tracking because they are more adapted to the context than Kalman filtering which imposes a linear state-space modelling and a necessarily white Gaussian noise. Preliminary results of these methods will be presented in this paper.