Aerial infrared target tracking in severe jamming using skeletal tracking technology

As various infrared countermeasures have been developed, aerial infrared target tracking is becoming increasingly challenging. In this paper, we develop an algorithm that can reliably track fighter in the case of severe infrared jamming. Since most parts of the target are obscured by a decoy, we propose an algorithm that can track a fighter by detecting its visible components rather than the whole target. Inspired by the human skeletal model of the Microsoft Kinect sensor, a skeletal model of a fighter comprising three components is designed. Then, a detection method based on Cascade R-CNN is used to detect those components. Situation assessment is made based on the detection result. As a result, the position of the target can be estimated through the detection result and skeletal model. We evaluated this proposed approach on an infrared image dataset against six state-of-the-art trackers. The experimental results demonstrate that our algorithm can reliably track a fighter in severe infrared jamming while running at a fast speed of 11.2 FPS.