Estimation of rotation in ISAR imaging based on local sharpness measure

The geometry of inverse synthetic aperture radar (ISAR) imaging suggests in principle a tomographic approach to the reconstruction of a 2D projection of an object illuminated by a radar signal. In theory this is straight forward, but the computational load of such reconstructions have in stead led to the use of fast Fourier transforms (FFTs) on a Cartesian grid. Using FFTs directly in ISAR imaging requires a small rotational motion to avoid blurring and defocusing. Alternatively, interpolation and polar reformatting can be applied prior to the reconstruction step. Also, time-frequency methods have been proposed to reduce the effects of blurring and defocusing of the reconstructed ISAR images. By applying tomographic reconstruction, the polar reformatting is the last stage of the ISAR imaging process, meaning that phase problems involved in interpolating complex data points are reduced. The objectives of this paper are; 1) To propose a method for estimating the rotational motion of an object based on maximizing the sharpness of a small region of the reconstructed image, and 2) To describe an efficient implementation of a tomographic ISAR image reconstruction in the case of non uniform rotation.