计算机应用研究2013，Vol.30Issue(6)：1910-1913,4.DOI:10.3969/j.issn.1001-3695.2013.06.082
基于CenSurE特征的虚实配准方法研究
New registration method based on CenSurE
胡佳 1汤光明 1孙怡峰 1刘佳2
作者信息
- 1. 信息工程大学,郑州450004
- 2. 解放军66075部队,山西大同037034
- 折叠
摘要
Abstract
For the high accuracy and real-time requirements of the registration in augmented reality,this paper proposed an adapting registration algorithm based on CenSurE feature.This method included offline initialization and online tracking.In the offline initialization stage,used the CenSurE-OCT filter and the MU-SURF descriptor for feature extraction and matching.Based on the set of matching features,it established the coordinate system and calculated the initial registration matrix.In the online tracking stage,it used an adaptive tracking method to obtain the correspondence relation between the feature points of current image and the reference points,and then used the error approximation method to obtain the registration matrix of current image.Experimental results show that the mean registration error of the method is 1.91 mm and the computing speed is 21 fps.Its comprehensive performance is better than the registration methods based on FAST,SIFT and SURF.This method has higher registration accuracy and better property of real-time.关键词
增强现实/虚实配准/CenSurE特征/特征匹配Key words
augmented reality / registration / CenSurE / feature matching分类
信息技术与安全科学引用本文复制引用
胡佳,汤光明,孙怡峰,刘佳..基于CenSurE特征的虚实配准方法研究[J].计算机应用研究,2013,30(6):1910-1913,4.
