Computer Engineering and Applications ›› 2021, Vol. 57 ›› Issue (20): 82-89.DOI: 10.3778/j.issn.1002-8331.2009-0263

Previous Articles     Next Articles

Improved Qtree_ORB Algorithm

YANG Shiqiang, BAI Lele, ZHAO Cheng, WANG Guodong, LI Dexin   

  1. School of Mechanical and Precision Instrument Engineering, Xi’an University of Technology, Xi’an 710048, China
  • Online:2021-10-15 Published:2021-10-21



  1. 西安理工大学 机械与精密仪器工程学院,西安 710048


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.

Key words: visual Simultaneous Localization And Mapping(SLAM), Oriented Fast and Rotated Brief(ORB), Qtree_ORB algorithm, uniformity



关键词: 视觉同时定位与建图算法(SLAM), ORB算法, Qtree_ORB算法, 均匀度