Improved small moving target detection method in infrared sequences under a rotational background

Appl Opt. 2018 Nov 1;57(31):9279-9286. doi: 10.1364/AO.57.009279.

Abstract

The infrared image sequence contains both the rotational background and a moving target, causing great difficulties in target detection and the occurrence of massive false alarm rates. This paper proposes an improved method using an optical flow field, which can estimate and compensate for the rotational motion background, thus detecting a small moving target. First of all, we select the Shi-Tomasi angular points for image detection and represent the background rotation with the sparse optical flow field of a complete image. Then we use the least squares method to estimate the parameters of the perspective model of background rotation so as to reconstruct the background motion vector field and compensate for background rotation. Finally we detect the small moving target with the differential method and the morphological treatment. The experimental results verify that this method can accurately detect a small moving target against the violently rotational background and provide an idea for the target interception algorithm under the condition that an infrared imaging seeker is overlooking.