An Evaluation Method of SAR Images Based on Kullback-Leiber Divergence

A quantitative evaluation method of synthetic aperture radar (SAR) simulation image based on the KK distribution and the Kullback-Leiber (KL) divergence is introduced in order to evaluate the validity of SAR templates by theoretical modeling. Firstly, the histogram of SAR image is modeled with the KK distribution. And then, the symmetric KL distance between two probability distributions is defined. By computing the KL distances between the simulated SAR images and the moving and stationary target recognition (MSTAR) SAR images of three typical ground vehicles (BMP2, BTR70, and T72), we conclude that the evaluation method based on the Kullback-Leiber divergence is effective.