SAR imagery from the perspective of multiscale Chirplet Transform

In SAR (Synthetic Aperture Radar) imagery, a major concern is related to finding a way of classifying the objects retrieved in the data acquisition made by a sensor. Dealing with complex-valued data requires finding an adequate descriptor that will provide enough information about the object captured and, in the end, will determine the type of the object. Taking into consideration the fact that the data acquisition model for SAR is based on chirp signals, an immediate choice for getting the desired features is the Chirplet Transform, which, practically, transforms the time space into a frequency-chirprate space. Therefore, the main purpose of the present paper consists in defining the multi-scale 2-D Chirplet Transform and finding a way of characterizing objects captured in a SAR image by using the method mentioned.