摘要
针对目前SLAM算法对特征提取的可靠性和鲁棒性要求越来越高的问题,基于ORB(二进制定向简单描述符)特征提取算法,提出了一种基于模块化区域分割的方法,解决了特征提取过程中提取无效特征点过多和特征点分布不均匀的矛盾。该方法使得ORB算法的特征提取更加均匀和合理,也更有利于图像的特征匹配。实验研究表明,该方法提高了算法的有效性和可靠性。
Aiming at the improvement of the reliability and robustness of the feature extraction in SLAM algorithm,an improved method of feature extraction is proposed. Based on the ORB(Oriented fast and Rotated Brief) feature extraction algorithm,an area segmentation method is proposed. It solves the problems of too more invalid feature points and the uneven distribution of feature points. This method makes the feature extraction of ORB algorithm more uniform and reasonable,and is more conducive to the image feature matching. The experimental results show that this method improves the effectiveness and reliability of the algorithm.
出处
《微型机与应用》
2017年第20期23-26,共4页
Microcomputer & Its Applications
关键词
ORB
模块分割
特征匹配
ORB
module segmentation
feature matching