摘要
双目立体视觉系统通过模拟人的双眼来获取三维世界的深度信息,该系统主要由摄像机标定、特征点提取、特征点匹配、三维坐标求取等4个模块组成。对于传统算法特征点提取速度与特征匹配不能满足实时性的要求,提出了基于ORB算法的双目视觉目标定位。该算法通过在Brief描述子上加入了旋转矩阵,使算法具备了旋转不变性以及有着很大的效率提升。实验结果表明,该算法特征点提取与匹配速度有着一个数量级以上的提高,大大增加了在双目视觉系统下的实时性。
Binocular stereo vision system through imitating human eyes to get three-dimensional world of depth information,system consists of camera calibration,feature point extraction and matching feature points and calculate the three-dimensional coordinates of four modules,to traditional algorithm and feature matching feature points extraction speed cannot meet the requirements of real-time,was proposed based on the ORB algorithm of binocular vision target,the algorithm described in Brief,son joined the rotation matrix,the algorithm with the rotation invariance and has a great efficiency,the experimental results show that the feature point extraction and matching speed of the algorithm has more than one order of magnitude increase,greatly increased under the binocular vision system real-time performance.
出处
《电子测量技术》
2017年第4期142-145,共4页
Electronic Measurement Technology
基金
河南理工大学博士基金(B2010-17)
河南省科技攻关项目(162102210228)
国家自然科学基金(U1504623)资助项目
关键词
双目视觉系统
ORB算法
目标定位
binocular stereo vision system
ORB algorithm
target location
