Infrared Small Target Detection Based on Isolation Constraint Under Complex Background

Abstract

In the development of infrared search and track (IRST) system, small target detection in complex background is a challenging task due to the low signal-to-noise ratio (SNR) of small targets and interference of irregular clutter. In this paper, a detection method based on isolation constraint is presented. Firstly, the suspected region with certain contrast is obtained by multilayer gray difference (MGD). Then we calculate the isolation index of the suspected region using the eigenvalues of Hessian matrix of the raw image to suppress the irregular clutter. In order to get accurate size of the target, a local scale estimation method is proposed. Experiment results demonstrate that the proposed method is accurate and robust, and superior to baseline methods in signal-to-clutter gain (SCRG) and receiver operating characteristic (ROC) curve.