Autofocus method based on successive parameter adjustments for contrast optimization

Abstract

A radar on a moving platform generates an initial synthetic aperture (SAR) image of a scene from a sequence of periodic pulse returns approximately motion compensated. The SAR image is formed from pixel intensities z(x,y) within a x,y extent of the initial synthetic aperture image. Targets are selected from the initial synthetic aperture image using a sliding window, computing a first entropy for the selected targets, and sorting the targets using the first entropy to obtain a target list having target elements, then concatenating the target elements to form a data matrix compatible in the azimuth dimension with a Fast Fourier Transform. A phase correction for autofocus is iteratively computed and applied to the initial synthetic aperture image using an inner loop, a mid loop and an outer loop. The phase correction is expressed using an orthogonal polynomial having a plurality n consecutive terms a, adenoting a quadratic term, and adenoting a last order term. The outer loop, using an L index, calculates an outer loop E(a) entropy for the quadratic term and an outer loop E(a) entropy for the last order term. Iterations within the outer loop continue until E(a)−E(a) is less than an outer loop tolerance. Similarly, the mid loop, and inner loop continue until the computation of their respective entropies meet a pre-set tolerance. The inner loop entropy uses a Golden Section search for computing the inner loop entropy.