A transformed image processing approach based on the continuous wavelet transform for non-dispersive scattering extraction

In this paper, we apply wavelet ideas to the stepped frequency backscattered data reflected from non-dispersive scattering centres of an object to obtain the transformed image in the range-scale domain. Each frequency component with a resolution matched to its scale can be analysed for the extraction of behaviour of the object to the incident high frequency wave in radar systems. The processing cuts the transformed image into many slices along the scale axis to produce multiresolution range profiles. Information about scattering centres are parameterised by a local modulus maxima algorithm in terms of scales and local peaks which presents the scattering distribution along the radial distance in the range-domain. To obtain reliable locations of the scattering centres, a number of multiresolution range profiles are used for the local maxima extraction. The coincident approach is then applied to identify exactly the locations of appropriated scattering centres. Feature vector sets derived from the significant range profile can be used for the automatic target recognition or target signature storage in the wideband stepped frequency radar.