Multiple-Target Tracking Framework for Aircraft in Airport Ramp Area
Veera V. Vaddi, Parikshit Dutta, Hui-Ling Lu, Jack Tsai · 2016
Current research develops a target-tracking framework that in turn enables a visionbased surveillance system concept suitable for airport ramp area operations. The surveillance approach is based on detecting and tracking aircraft only using cameras. A multiple target tracking algorithm based on nearest neighbor standard filter is used in the research. The measurement model consists of the position and velocities of the aircraft. The aircraft positions are obtained from an object detection algorithm and the velocities from optical flow estimation. Data association between measurements and targets is achieved by solving a linear assignment problem. An innovative aspect of this research involves the usage of Hausdorff metric to quantify the distances between established targets and measurements. A linear model with additive Gaussian noise is assumed for target propagation and Kalman-filter is used to remove noise and localize the target. Moreover, a novel methodology for target death and birth is proposed which is consistent with the current target tracking framework. We apply the proposed multiple-target tracking algorithm to track aircraft in a cluttered ramp area environment of an airport. It is observed that the target tracking algorithm is able to consistently track aircraft through different frames.