HFSWR Clutter Mitigation: from Wavelets to Empirical Mode Decomposition

Maritime surveillance of the Exclusive Economic Zone (EEZ) is a present military and civilian challenge. The High Frequency Surface Wave Radar, as its coverage range is not limited by the radio horizon, is well-suited to fulfil this task. HFSWRs are based on the ability of HF waves (3 MHz to 30 MHz) to propagate along the earth curvature: it is possible to detect targets up to few hundred kilometres. Effect of ionospheric clutter and sea clutter can, however, strongly limit target detection. Ionospheric clutter results from disturbance in ionization while sea clutter is caused by roughness of sea surface [1]. In previous works we proposed using wavelets to remove the clutters from high frequency surface wave radar data [1] [2]. We have also considered the issue of the best decomposition basis. Indeed, both of clutters are the radar signatures of random process [3], and we wondered if an evolving basis could overcome the issue. Empirical Mode Decomposition (EMD) might offer such a possibility. Moreover, EMD and wavelets seems to be connected [4]. Thus, we have tried to carry out an EMD-based clutter mitigation. The initial idea is still to use multi-scale analysis to turn to good account the differences in variation scales of targets, ionospheric clutter and sea clutter. Apart from the fact that basis functions (more precisely the generatrix family) is self-determined by the EMD. We have looked into using complex bi-dimensional EMD, morphological operators and Graham-Schmidt orthogonalization. We are presenting here the first results of our investigations.