Please choose your delivery country and your customer group
In classical high-frequency radar imaging, like synthetic aperture radar imaging, man-made target contributions are often well described by decomposing their signature into a set of bright points. The basic model supposes that these elementary reflectors are independent of the relative angular aspect and of the observation frequencies. Many feature extraction methods, such as CLEAN/RELAX-based algorithms, are built on this hypothesis and so it is currently used in assisted/automatic target recognition algorithms (ATR). However, this simple model cannot describe the variability of the signatures one can observe in image databanks. The authors propose extending the target feature extraction capacities of the CLEAN/RELAX algorithm to dispersive scatterers using generalised wavelets.