序列全景图像的特征提取与匹配

被引:4
作者
高明 [1 ]
曹洋 [2 ]
方帅 [1 ]
机构
[1] 合肥工业大学计算机与信息学院
[2] 中国科学技术大学自动化系
关键词
虚拟柱面; 非线性拟合; 尺度空间; Harris; Laplace;
D O I
暂无
中图分类号
TP391.41 [];
学科分类号
080203 ;
摘要
为了提高SLAM系统特征提取和匹配的稳定性,文章提出了在全景展开图像的多尺度空间上提取特征的方法。该方法首先对全景图像进行基于虚拟柱面的展开和矫正,然后在此基础上为展开图像建立多尺度空间,最后在多尺度空间内通过Harris特征跟踪器来选择候选特征点,并用Laplace公式选择出在尺度空间范围内具有局部最大值的特征点。这些特征点具有尺度、旋转和光照不变性,并且在有限视角变化下也能保持稳定性。实验证明,这种方法比在未展开图像或展开图像的单尺度空间内提取特征更具稳定性。
引用
收藏
页码:449 / 452+460 +460
页数:5
相关论文
共 12 条
[1]  
Indexing based on scale invariant interest points. Mikolajczyk K,Schmid C. Proceedings of the8th International Conference on Com-puter Vision . 2001
[2]  
Information samplingfor opti-mal i mage data selection. Winters N,Santos-Victor J. Proceedings of the Ninth In-ternational Symposium on Intelligent Robotics Systems . 2001
[3]  
Omni-directional vi-sion for robot navigation. Winters N,Santos-Victor J,Caspar J. Proceedings of the 1st Inter-national IEEE Workshop on Omni-directional Vision atCVPR . 2000
[4]  
SLAM with omni-directional stereovision sensor. Ki mJ H,Chung MJ. Proceedings of the International Confer-ence on Intelligent Robots and Systems . 2003
[5]  
SLAM with panoramic vision. Lemaire T,Lacroix S. Journal of Field Robotics . 2007
[6]  
Comparing and evaluating interest points. C. Schmid,R. Mohr,C. Bauckhage. Proceedings of the 6th International Conference on Computer Vision . 1998
[7]  
Rover localization in natural environments by indexing panoramic image. J. Gonzalez-Barbosa,,S. Lacroix. Froc of the International Conference on Robotics and Automation, IEEE . 2002
[8]  
Vision-based navigation and environmental representations with an omnidirectional camera. Gaspar J,Winters N,Santos-Victor J. IEEE Transactions on Robotics and Automation . 2000
[9]  
Omnidirectional visual navigation. Winters N,Santos-Victor J. In proc. of the 7th Int . Symp. on Intelligent Robotic Systems (SIRS’ 99)[C], Coimbra, portugal . 1999
[10]  
Object recognition fromlocal scale-invariant features. Lowe David G. Proceedings of the International Conference on Computer Vision . 1999