基于单目视觉的实时目标距离测量

利用计算机视觉测量目标距离的方法广泛应用于机器人、军事目标跟踪、自主导航、精密检测等系统当中。研究和设计了一种基于单目视觉的目标距离测量的方法,实现了实时目标距离测量。首先,利用黑白棋盘格对单目摄像机进行标定,得到摄像机的内参,通过内参对目标模型注册,生成含有3D坐标、2D坐标、特征点以及ORB描述子的特征匹配文件。然后,通过RANSAC算法从匹配点中随机选取一些特征点,使用Pn P算法计算目标距离。通过实验,验证了该方法可以有效地对三维空间中的移动目标进行有效的三维距离测量。

In this paper,a method of fixed target distance measurement based on monocular vision is studied and designed,and real-time measurement of target distance is realized.First of all,the black and white checkerboard on the monocular camera calibration is applied to get the camera's internal parameters,which are used to register the target model to generate a feature matching file including 3D coordinates,2D coordinates,feature points and ORB descriptor.Then,a good matching point is randomly selected by RANSAC algorithm,and the target distance is calculated by PnP algorithm.