A guide head airborne target detection and tracking algorithm in the ground-to-air context

In this paper, a complete set of algorithms including filtering, detection, and tracking is proposed for the challenge of airborne target detection and tracking of the guide head in the ground-space context. In the filtering stage, the algorithm first realizes the detection of sky and ground regions by Sobel operator edge detection and Hough transform, and then uses high-frequency filtering and phase difference method to filter the sky and the ground respectively to suppress the background noise. In the detection phase, feature points are extracted by combining curvature filtering and calculating the saliency features (SOT), and effective targets are filtered by trajectory correlation and filtering. In the tracking stage, the phase correlation method is utilized to track the target, and the template update and window size adaptive mechanism are used to adapt to the target size and morphology changes, and finally the aiming point is calculated by the center of mass. The experimental results show that the algorithm can effectively eliminate the background clutter interference, and can effectively accomplish the target detection and tracking task under the complex ground and air background.