基于微型飞行器机载视觉的目标识别与状态估计方法

Airborne visual object identification and status estimation method based on micro-flight vehicle

为了实时获得微型飞行器着陆时与着陆平台的相对位置关系,本文提出了一种基于微型飞行器机载视觉的目标识别与状态估计方法。该方法利用曲面拟合方法获得着陆平台图像边缘的梯度方向和亚像素坐标,并将其作为匹配信息进行相似度运算,匹配成功后再通过运动方程模拟移动平台的运动特性,利用卡尔曼滤波对移动平台下一时刻的位置进行预测。最后进行偏移量的计算,将实验所得的数据通过几何变换可以得出飞行器与着陆平台的相对位置参数。实验结果表明该算法能够实时准确计算出飞行器与视觉着陆标识之间的相对位置关系。

In order to obtain the relative position relation between the landing platform and the micro air vehicle (MAV) when it lands in real time, a method of target recognition and state estimation based on the airborne vision of the micro-aircraft is put forward. The gradient direction and sub-pixel coordinates of the image edge of landing platform are obtained by using curved surface fitting method, which are used as matching information for similarity calculation. After successful matching, the motion characteristics of the mobile platform are simulated through the motion equation, and the position of the mobile platform is predicted by using Kalman filter. Finally, the offset is calculated. The experiment data can be used to calculate the position parameters of the aircraft and landing platform through geometric transformation. Experiment results show that the proposed algorithm can calculate the relative position relation between the aircraft and the visual landing identification in real time.