摘要
在视觉SLAM中,特征点的提取和准确的特征匹配对机器人的位姿推断具有重要作用。针对传统ORB算法特征点分布不均匀,容易出现簇集的问题和Qtree_ORB算法特征点过均匀等问题,提出了一种基于四叉树改进的ORB特征提取算法。对每层图像金字塔进行自适应网格划分,采用自适应阈值来进行特征点提取;根据每层图像金字塔所提取特征点数目对四叉树的划分深度进行限制,减少冗余特征点;设定最小阈值来减少低质量特征点的提取;在Mikolajczyk数据集上对改进算法的均匀度和匹配性能进行测试,在TUM数据集上对改进算法在ORBSALM2系统中的精度进行测试。结果表明改进算法能够有效提高其均匀度,并且保持着较高的匹配精度;在ORBSLAM2测试中,改进算法能有效改进SLAM系统的轨迹精度和漂移程度。
In visual SLAM,feature point extraction and accurate feature matching play an important role in robot pose inference.Aiming at the uneven distribution of feature points in the traditional ORB algorithm,the problem of clustering and the over-uniform feature points of the Qtree_ORB algorithm,an improved ORB feature extraction algorithm based on quadtree is proposed.Each layer of the image pyramid is adaptively meshed,and adaptive thresholds are used for feature point extraction.It limits the division depth of the quadtree according to the number of feature points extracted from each layer of the image pyramid to reduce redundant feature points.It sets a minimum threshold to reduce the extraction of low-quality feature points.The uniformity and matching performance of the improved algorithm are tested on the Mikolajczyk dataset,and the accuracy of the improved algorithm in the ORB-SALM2 system is tested on the TUM dataset.The results show that its uniformity and matching accuracy can be effectively improved by the improved algorithm.In the ORB-SLAM2 test,the trajectory accuracy and drift degree of the SLAM system can be effectively improved by the improved algorithm.
作者
杨世强
白乐乐
赵成
王国栋
李德信
YANG Shiqiang;BAI Lele;ZHAO Cheng;WANG Guodong;LI Dexin(School of Mechanical and Precision Instrument Engineering,Xi’an University of Technology,Xi’an 710048,China)
出处
《计算机工程与应用》
CSCD
北大核心
2021年第20期82-89,共8页
Computer Engineering and Applications
基金
国家自然科学基金(51475365)
陕西省自然科学基础研究计划项目(2017JM5088)
陕西省教育厅重点实验室科研计划项目(13JS070)。
