부산대학교 도서관

Flying target identification is an important and challenging problem in airborne surveillance systems. In this report, we propose a method to automatically identify the flying targets by physical information (coordinates, heading, speed), time, and identification information (3/A code). This method includes two steps: features extraction and building a machine learning model. In the features extraction step, features that are extracted include cell indexes corresponding to coordinates, information of flight path, time in day/night format, heading, speed, and 3/A code, constructing n-dimensional vector. This vector is used as input for training a Random Forest model, to automatically identify class labels of flying targets.