Video-based detection of subpixel no-reflecting objects in the presence of fluctuated background

Abstract

This work presents a novel generalized likelihood ratio (GLR) methodology to design a detection algorithm for a sequence of images in the presence of Gaussian structured background and noise. We focus on the detection of a subpixel target both with small target's energy and without it (dark target). It is assumed that the background power is known a priori under the null hypothesis and is unknown under the alternative hypothesis. This situation is typical in the case of subpixel detection. We derive the GLR test (GLRT) for this problem. The designed detector is theoretically justified and numerically evaluated. The structure of the designed detector consists of two detectors. The first detector is the known matched subspace detector and the second is the background's power change detector. The second detector help to detect the black subpixel targets (without reflected energy). Both the theoretical and computer simulation results have shown that the proposed detector outperforms the conventional one.