Autofocus is a key step of inverse synthetic aperture radar (ISAR) imaging.
In this paper four new approaches to autofocussing based on the applicatio
n of beamforming and subspace concepts to ISAR imaging are developed. Their
relations to maximum likelihood (ML) estimation are identified. A common f
eature of these techniques is the estimation of the complex vector formed b
y the exponential function of phase rather than phase itself so that phase
unwrapping is obviated. The Cramer-Rao lower bound (CRLB) of the estimated
complex vector corresponding to translational motion and the CRLB of the es
timated distance between two scatterers are derived. The results of process
ing simulated and real data confirm the validity of proposed approaches. (C
) 2001 Elsevier Science B.V. All rights reserved.