Parametric scatterer extraction method for space‐target inverse synthetic aperture radar image CLEAN

Abstract

Owing to the time variance of the inverse synthetic aperture radar (ISAR) imaging plane of space targets, scatterers with large heights relative to the imaging plane migrate through several Doppler cells. This defocussing in the azimuth direction of each scatterer cannot be compensated via the traditional migration through the resolution cell (MTRC) compensation method because the height of each scatterer is quite different and independently distributed in the ISAR image. In this study, a parametric azimuth-defocussing model-based CLEAN method is proposed for the space-target ISAR image. First, highorder defocus parameters and their influences on the image of space targets were analysed at different arc segments. Subsequently, a mathematical azimuth-defocussing expression of the ISAR image scatterers was derived after MTRC and azimuth high-order phase compensation, and an approximate defocus expression based on the Fresnel integral was provided for computational efficiency. This model can estimate the complex amplitude, coordinates and defocus parameters of each scatterer in the image domain by applying a sequential quadratic programing algorithm. Finally, the validity and stability of the proposed algorithm for the isolated and coupled scatterer extraction were verified at different signal-to-noise ratios, and different defocus parameters were verified through simulations. K E Y W O R D Sinverse synthetic aperture radar (ISAR), parameter estimation, radar imaging, radar resolution | INTRODUCTIONAn inverse synthetic aperture radar (ISAR) uses a synthetic aperture formed by the relative rotation between the target and radar to obtain a high-resolution image of the target. Owing to its all-weather imaging capability, it has been widely used in military and civilian fields. To improve the imaging quality or extract more information, spatially variant apodisation [1-3], relaxation (RELAX) [4][5][6][7], and other methods are commonly used to enhance images.The CLEAN algorithm was first proposed by astronomers [8] and was later adopted in radar imaging to reduce the interference of sidelobes and speckle artefacts [9]. In recent years, the CLEAN algorithm has been widely used in ISAR imaging. As a deconvolution signal-processing technique, the CLEAN algorithm is usually used for feature extraction of all scatterers in an image [10], wherein it iteratively estimates the point-spread function (PSF) of the strongest scatterer and subtracts it from the image. The PSF of the scatterers can be assumed to be an ideal two-dimensional (2D) sinc function when its high-order phase has been compensated. However, this assumption fails in many realistic scenarios. An accurate PSF mathematical expression model is a prerequisite for fast and successful implementation of the CLEAN algorithm. Owing to the differences in imaging targets, pre-processing methods and processing domains in which the CLEAN method is executed, PSF model variations exist [11][12][13][14][15][16][17][18][19]. The CLEAN algorithm in Ref. [11] extracts parameters ...