Please wait a minute...
J4  2009, Vol. 43 Issue (5): 855-859    DOI: 10.3785/j.issn.1008-973X.2009.05.013
自动化技术、计算机技术     
Harris尺度不变性关键点检测子的研究
程邦胜1,2,唐孝威1
(1.浙江大学 物理系交叉学科实验室,浙江 杭州 310027; 2.浙江大学 生物医学工程学系,浙江 杭州 310027)
Study of Harris scale invariant keypoint detector
CHENG Bang-sheng1,2, TANG Xiao-wei1
(1. Interdisciplinary Laboratory of Department of Physics, Zhejiang University, Hangzhou 310027, China;
2. Department of Biomedical Engineering, Zhejiang University, Hangzhou 310027, China)
 全文: PDF(1335 KB)   HTML
摘要:

在特定的参数设置下Harris尺度不变性检测子不能提供足够数目的稳定关键点,以往研究据此断定Harris尺度不变性检测子不稳定,不是有效的特征检测子.在构造Harris角点值的尺度空间过程中, 存在一系列参数影响着Harris角点值在尺度空间中极值点的数目,从而决定了Harris尺度不变性检测子所能获取的稳定关键点的数目.对这个参数空间进行了系统研究,发现积分尺度与微分尺度的比值对Harris尺度不变性检测子能否检测到足够数目的稳定关键点具有决定性的影响.当这一比值减少时,Harris尺度不变性检测子所获取的稳定关键点的数目快速增长;当这一比值小于0.8时,Harris尺度不变性检测子所能获取的稳定关键点的数目开始超出DoG检测子所能获取的稳定关键点的数目.这个有效的参数区间大大增加了Harris-Laplace检测子所能获取的稳定关键点的数目.与Harris-Laplace检测子以及DoG检测子进行比较,具有有效参数的Harris尺度不变性检测子具有最佳的稳定性能,这个结果纠正了关于Harris尺度不变性检测子不稳定的错误结论.

Abstract:

Previous works demonstrated that Harris scale invariant detector with particular parameters cannot extract enough stable keypoints from images, and this led to the  conclusion that Harris scale invariant detector was unstable and invalid for feature detection. In the process of constructing the Harris cornerness scale space, there are some  parameters influencing the number of local maxima in the scale space and determining the number of keypoints that can be extracted by Harris scale invariant detector. This work  systematically investigated this parameter space and found that the ratio between integrate scale and differentiate scale had deterministic influence on the number of keypoints  that could be extracted by Harris scale invariant detector. When decreasing this ratio, the detected keypoint number increases rapidly; when this ratio equals to 0.8, the  detected keypoint number is comparable to that detected by DoG detector. This valid parameter interval greatly increases the keypoint number detected by Harris-Laplace detector.  When comparing with DoG detector and Harris-Laplace detector, Harris scale invariant detector obtains the best performance. This result corrects the previous conclusion about  the instability of Harris scale invariant keypoint detector.

出版日期: 2009-11-18
:  TP242.6  
通讯作者: 唐孝威,男,教授,院士.     E-mail: fmrilab@zju.edu.cn
作者简介: 程邦胜(1972-),男,安徽安庆人,博士生,从事计算机视觉、视觉信息处理研究.
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  

引用本文:

程邦胜, 唐孝威. Harris尺度不变性关键点检测子的研究[J]. J4, 2009, 43(5): 855-859.

CHENG Bang-Qing, TANG Xiao-Wei. Study of Harris scale invariant keypoint detector. J4, 2009, 43(5): 855-859.

链接本文:

http://www.zjujournals.com/eng/CN/10.3785/j.issn.1008-973X.2009.05.013        http://www.zjujournals.com/eng/CN/Y2009/V43/I5/855

[1] SCHMID C, MOHR R, BAUCKHAGE C. Evaluation of interest point detectors [J]. International Journal of Computer Vision, 2000, 37(2): 151172.
[2] HARRIS C, STEPHENS M. A combined corner and edge detector [C]∥ Alvey Vision Conference. Manchester: [s.n.], 1988: 147151.
[3] LINDEBERG T. Edge detection and ridge detection with automatic scale selection [J]. International Journal of Computer Vision, 1998, 30(2): 117156.
[4] LINDEBERG T. Feature detection with automatic scale selection [J]. International Journal of Computer Vision, 1998, 30(2): 79116.
[5] MIKOLAJCZYK K, SCHMID C. Indexing based on scale invariant interest points [C]∥ Proceedings of the 8th IEEE International Conference of Computer Vision. Vancouver:  IEEE, 2001: 525531.
[6] MIKOLAJCZYK K, SCHMID C. Scale & affine invariant interest point detectors [J]. International Journal of Computer Vision, 2004, 60(1): 6386.
[7] LOWE D. Object recognition from local scale-invariant features [C]∥ Proceedings of the 7th IEEE International Conference on Computer Vision. Corfu: IEEE, 1999: 11501157 .
[8] LOWE D. Distinctive image features from scale-invariant keypoints [J]. International Journal of Computer Vision, 2004, 60(2): 91110.
[9] MIKOLAJCZYK K, TUYTELAARS T, SCHMID C, et al. A comparison of affine region detectors [J]. International Journal of Computer Vision, 2005, 65(1/2): 4372.

[1] 谷雨, 李平, 韩波. 基于分层粒子滤波的地标检测与跟踪[J]. J4, 2010, 44(4): 687-691.
[2] 蒋荣欣, 张亮, 田翔, 陈耀武. 多机器人队形变换最优效率求解[J]. J4, 2010, 44(4): 722-727.
[3] 奚海燕, 牟同升, 李俊凯, 等. 平板显示器彩色运动伪像的测量与评价[J]. J4, 2009, 43(6): 1158-1162.