DEVELOPMENT OF AN ISAR IMAGING ALGORITHM

Abstract

A radar image is the spatial distribution of reflectivity corresponding to the object. An essential requirement for producing a high-quality image is the system resolution, both in range and azimuth directions. In this project, an Inverse Synthetic Aperture Radar (ISAR) imaging algorithm has been developed for the purpose of improving imaging resolution. This imaging algorithm is different from the conventional methods. The signal format of this new algorithm is a stepped-chirp train instead of the usual Linear Frequency Modulation (LFM) signals. This signal waveform increases the ranging accuracy and nominal resolution in range direction, which are improved not by increased overall bandwidth but via a novel algorithm, namely the vernier algorithm, which employs two Fast Fourier Transform (FFT) processing in the range direction. In this method, the range profile of a target along various rotated positions is first obtained. Two procedures are proposed. The first procedure is for targets which rotate through only a small angle (of 5° or less) during the imaging process. The second procedure, on the other hand, has done away with most of the constraints imposed in the first method and is capable of imaging rapidly rotating targets. Comparison between the proposed algorithms and the conventional Range-Doppler technique has also been made. When a target moves along a straight line, the effective rotation is accompanied by undesirable translational motion which must be compensated. Among several motion compensation methods, the scheme of scattering centroid tracking is chosen to take advantage of the high ranging accuracy of the vernier method. A second method, the Wigner-Ville Distribution (WVD) analysis is also investigated. The WVD method is modified to take advantage of the processing method of stepped-chirp pulse. After the first FFT in the Vernier method range processing, the Wigner-Ville distribution is used to estimate the phase change history of the target rotation center from the received signal along the cross-range direction without the needs to track the scattering centroid in the space domain. Comparisons show that the proposed method is superior to the conventional centroid tracking method. Its performance in the presence of white Gaussian noise is also investigated.