基于视觉的自主空中加油锥套检测与跟踪

Drogue Detection and Tracking Based on Vision for Autonomous Aerial Refueling

加油锥套检测与跟踪是无人机自主空中加油得以实现的前提,提出一种锥套检测算法,通过添加人工标记,结合颜色通道相减的方法对锥套标记进行提取,利用最大类间方差法进行阈值分割,结合中值滤波、形态学操作完成锥套检测。对原始图像进行噪声、对比度、亮度干扰,验证检测算法的鲁棒性。在该检测算法的基础上,结合上一帧锥套的图像尺寸和锥套运动速度确定锥套的ROI(region of interest)区域,提出一种锥套跟踪算法,在ROI区域内完成对锥套的检测和定位。仿真实验结果表明,引入跟踪后,帧处理速度大大提高,能很好的满足实时性需求。

As drogue detection and drogue tracking is the base of autonomous aerial refueling. In this paper, a drogue detection method is proposed by add markers. The rough drogue markers detection is obtained by color channel subtraction. The Otsu method is used to threshold the rough drogue image. Then the complete drogue region is implemented by median filtering and morphological operations. The noise, contrast, and brightness is added on the original image in order to verify the robustness of the detection algorithm. Based on the drogue detection algorithm, a drogue tracking method is proposed with the use of the drogue image size and drogue velocity between two frames, the tracking region of interest （ROI） area is obtained. Drogue detection and location will be conducted in the ROI area. The simulation results show that the frame processing speed will greatly improve after the tracking algorithm of this paper is used, which can meet real time requirements.