无人机多机动目标自主探测与跟踪

Autonomous detection and tracking of multiple maneuvering targets for unmanned aerial vehicles

中文关键词  无人机  多目标  自主探测  关联滤波  目标跟踪

英文关键词  unmanned aerial vehicles  multiple targets  autonomous detection  association filtering  target tracking

中文摘要

      无人机机载相机图像中机动目标尺寸较小而且会发生显著变化, 加上大量的背景噪声干扰, 给目标探测和 跟踪带来很大困难. 针对这些问题, 本文提出了一种在无人机机载相机图像序列中自主探测与跟踪多个机动目标的 方法. 首先, 提取目标的图像数字特征并采用级联分类算法进行特征分类, 得到目标的强分类器, 对目标进行自主 探测搜索. 然后, 基于全局最优关联算法对探测回波进行关联滤波, 实现对多个机动目标的跟踪与识别, 其中最优 关联代价矩阵融合了距离和方向信息, 提高了关联和跟踪的鲁棒性. 将无人机航拍图像序列中的地面坦克作为目 标进行实验, 结果表明本文算法可以实现对多个机动目标的自主探测和跟踪, 并具有较好的跟踪鲁棒性.

英文摘要

      The interested targets in UAV (unmanned aerial vehicles) aerial image sequences are always small in size, and have significant changes in appearance when coupled with a lot of background noises, making the target detection and tracking a very difficult task. To solve these problems, this paper presents a method to autonomously detect and track multiple maneuvering targets in UAV aerial image sequences. Firstly, image features of the targets are extracted and classified through cascade classification algorithms to obtain the strong classifier of the targets; then, the targets can be autonomously detected by using the classifier. The detection results are associated and filtered through global optimal association algorithms to achieve the purpose of tracking and recognition of multiple maneuvering targets. The optimal association-cost matrix is simultaneously fused with distance and direction information which can improve the tracking robustness. Using the ground tanks in UAV aerial image sequences as the interested targets, the experimental results showed that the proposed method can autonomous detect and track multiple maneuvering targets with desired tracking robustness.