火炮身管指向的视觉测量方法

Vision measurement for the pose of cannon barrel

为解决自行高炮火控系统精度测试过程中的身管指向(高低角和方位角)这一难题,提出一个基于计算机视觉技术的非接触测量方法。测量前,在火炮身管上固定一个棋盘状平面模板作为测量标志物。首先,将Harris算法提取到的棋盘板上的角点坐标进行相机成像模型的解算(即相机标定),结合扩展卡尔曼滤波(EKF)算法得到相机的内参。然后,提取待测位置图像的角点,运用最小二乘拟合算法解出该位置的外参矩阵。最后,根据身管的运动模型以及标志物与之的相对位置关系,推导高低角、方位角与外参矩阵满足的约束方程,进而解出这两个角度。实验结果表明,本文方法的测量精度达到了0.1°以内,基本满足实际需求。

The pose measurement of cannon barrel （vertical angle and azimuth angle） is difficult but im- portant in the precision analysis of SPAAG fire control system. To solve this problem, this paper pres- ents a contactless measuring method based on computer vision. Before measurement, a checked planar faceplate is fixed on the cannon barrel as a marker. Firstly,the coordinates of X corner points are got u- sing Harris cornerd detection algorithm to calibrate the camera, and the camerars intrinsic parameters are obtained based on extended Kalman filter. Secondly, the corner points are extracted from the image of test position, and then the extrinsic parameter matrix of the corresponding position is ealculatted combi- ning with the least square algorithm. Finally, according to the motion model of cannon barrel and its posi- tion relative to the marker, the constraint equations between extrinsic parameter and vertical angle and azimuth angle are derived, and then the two angles are figured out. Experiments validate this method, and the results indicate that the measuring precision of this method is less than 0.1°.