关键词: 无人机编队, 贪婪算法, 轮廓提取, 多目标跟踪, 目标遮挡

Abstract: Aiming to meet the requirements of multiple targets accurate tracking of UAV formation video sequences, STK-3D modeling software was utilized to simulate the UAV formation flight video and capture the moving regions by means of double difference operator and multi-resolution connectedness label operator on the simulation video. By using the rectangular moving regions as an initial contour, an improved greedy algorithm was proposed to realize the accurate contour drawing of the UAV targets. The center position of the moving regions being an input variable, the on line Kalman filter group were established to track each target in the video. Consequently, the way was discussed on how to determine and segment the objects when occlusions occur. The simulation results show the efficiency.

Key words: UAV formation, greedy algorithm, contour drawing, multiple targets tracking, object occlusion