Time-Frequency Analysis applied to High Resolution SAR Images

Abstract:
The radar backscattering features of singular targets critically depends on observation angle and chirp frequency. Such variations cannot be appreciated on standard SAR images (> 3 m resolution), as both integration time and chirp bandwidths are reduced. However, the resolution improvement will allow for the generation of numerous non-stationary effects. This paper assesses their detection and analysis on such imagery (< 1 m resolution), based on sub-band decomposition or time-frequency techniques. The sub-band decomposition process is combined with similarity measurement techniques such as Mutual Information (MI) or Internal Hermitian Product (IHP). The time-frequency transform is based on a Bertrand Class implemented by a wavelet filter. Applications of the introduced time-frequency techniques include SAR synthesis optimisation or target identification applications.