Time-frequency tomographic imaging of a rotating object in a narrow-band radar

Abstract

The backscatter from radar object carries Doppler information of scatterers on the object determined by the radial velocity of scattering points and the radar transmitted frequency. For a rotating object this information is contained in the frequency characteristics over varying aspect angle. Frequency characteristics are used to create projections for Doppler radar tomographic imaging. This paper presents a method for high resolution imaging of a rotating target using a time-frequency transform of a returned signal as tomographic projections. The resolution of a tomographic image depends not only on radar system parameters but also depends on the resolution of input projections. The reassigned spectrogram is proposed for building of tomographic projections, due to its possibility of squeezing of frequency spread. The reassigned spectrogram is sensitive to noise so the denoising procedure in the time-frequency domain must be performed before the reassignment procedure. The denoising is performed by removing Short Time Fourier Transform (STFT) noise coefficients below the appropriate threshold. The STFT is a linear time-frequency transform and coefficients, which belong to the signal and coefficients which belong to noise can be analyzed separately. The efficiency of the proposed idea of imaging is supported by results of numerical experiments.