Ship Detection in SAR Images Based on Lognormal $\rho$ -Metric

Abstract

Information geometry emerged from the study of the geometrical structure of a manifold of probability distributions. It defines a Riemannian metric uniquely, which is Fisher information metric. However, the classical Fisher metric method has a limitation that it does not overcome the problem of inhomogeneous and nonstationary clutter for ship detection in synthetic aperture radar (SAR) images. By combing lognormal model and the Riemannian geometry, this letter presents a modified Fisher metric (lognormal ρ-metric) based on information geometry. Experiments show that lognormal metric for ρ in the ship detection from SAR images can be tuned to increase the performance of improving contrast between the object and background, and reducing false alarms.