Flash and Persistent Point Scatterer Feature Extraction from Radar Images

The talk will discuss the modelling of flashes buried within radar range compressed, or raw data. It will be shown how variants of the CLEAN algorithm can speedily efficiently extract all the flashes. Persistent point scatterers are regarded here as a special case of flashes, of long duration in the time/angle domain. A `flash representation' of a target closely related to its ISAR image is shown. The talk will discuss efforts to validate the flash representation against flashes direct from CAD models, or from electromagnetic prediction codes such as QinetiQ's SPECTRE. Implications for feature-based ATR, scatterer interpretation, and human interpretability of radar images will be discussed