基于特征匹配算法的双目视觉测距

距离测量作为障碍物检测以及路径规划的前提和基础是机器人研究领域的一个重要分支。在众多测距方法中,由于双目立体视觉具有信息丰富、探测距离广等优点被广泛应用。本文将改进的SIFT特征匹配算法应用到双目视觉测距与标定系统中。首先建立双目视觉测距模型,测量值由空间物点在左右摄像机下的像素坐标值决定;其次根据该模型的特点提出了基于平行光轴的双目立体视觉标定方法;最后利用改进的SIFT特征匹配算法,提取匹配点的像素坐标完成视觉测距。实验结果表明,根据测量数据对障碍物进行三维重建,相对距离与真实场景基本吻合,能够有效地指导机器人进行避障。

Binocular stereo distance-measurement method based on feature matching algorithm

Distance measurement is the prerequisite and basis of the obstacle detection and path planning.It is an important branch of the robot research area.Ranging in many ways,binocular stereo vision with rich information,wide detection range and many other advantages has obtained widespread application.In this paper a binucolar stereo distance-measurement and calibration method based on SIFT feature matching algorithm is presented.First,a model of depth measurement based on stereo vision is built,the measurement result was decided by the pixel coordinate of space points.Secondly,according to the characteristics of the model,the binocular stereo vision calibration method based on parallel optical axis is proposed.Finally,the pixel coordinates of matching points are extracted with improved SIFT feature matching,and measurement distance is completed.Experimental results show that three-dimensional reconstruction can guide the robot obstacle avoidance effectively.