Hierarchical Wavelet Representations of Ship Radar Returns

Abstract : High-resolution ship radar returns contain in their structure substantial information about the target that can be used to better identify complex targets consisting of many scatterers. This applies to many forms of radar signatures, including the amplitude of pulsed radar (PR) returns, the phase of pulsed radar returns, Doppler radar (DR), synthetic aperture radar (SAR) returns, inverse synthetic aperture radar (ISAR) returns, and millimeter-wave (MM-wave) radar returns. With the increasing resolution of modern radars it is at least theoretically possible to store many of the possible returns (i.e., returns organized according to aspect, evaluation, and pulsewidth) of a complex target and use them in the field for target identification. This is true for naval targets (e.g., ships and submarines) in particular. The advantage of the increasing radar resolution is the availability of more detailed information, and ultimately of specific features, characteristic of the radar return from a specific ship. The disadvantage is that these very detailed characteristics require an ever-increasing amount of computer memory for storage. The latter not only results in unfeasible memory requirements but it also slows down the search time in real field operations. It is therefore important to develop extremely efficient ways of compressing the representations of high-resolution data returns from real ships, and to design efficient coding schemes that operate in a hierarchical manner on the compressed representations to recover the ship identity. It is our contention that multiresolution representations of the radar data, followed by properly designed hierarchical clustering, are key means of achieving both objectives.