YAN Wen-long,ZHU Yi,LU Jun.A Visual Tracking Algorithm Combining ORB Feature and Mean Shift[J].Journal of Chengdu University of Information Technology,2017,(06):613-617.[doi:10.16836/j.cnki.jcuit.2017.06.008]
一种融合ORB特征和Mean Shift的视觉跟踪算法
- Title:
- A Visual Tracking Algorithm Combining ORB Feature and Mean Shift
- 文章编号:
- 2096-1618(2017)06-0613-05
- 关键词:
- 视觉跟踪; Mean Shift; ORB特征; 嵌入式平台; 实时性
- Keywords:
- visual tracking; Mean Shift; ORB feature; embedded platform; real-time
- 分类号:
- TP391.4
- 文献标志码:
- A
- 摘要:
- 由于嵌入式系统对于运算的实时性要求较高,一些较为成熟的计算机视觉跟踪算法不能很好地满足这一要求。为解决这一问题,介绍一种结合ORB(oriented fast and rotated brief)特征点和Mean Shift的算法用以视觉跟踪。传统的Mean Shift算法运行速度较快,但在目标被遮挡下容易失效,导致跟踪结果不理想,所以决定采用融合ORB 特征检测算法的Mean Shift算法来实现目标跟踪。该算法通过检测目标的初始位置,并根据与模板匹配的特征点计算出目标在当前帧的尺度以及旋转角度,从而提高搜索窗口的精度。通过对比实验验证了:与传统的Mean Shift及其改进算法相比,文中介绍的算法在跟踪物体部分被遮盖时也显示出良好的健壮性,并且在跟踪的实时性上有良好的表现。
- Abstract:
- As the real-time requirements of the embedded system for computing are critical, some of the traditional vision tracking algorithm can hardly meet them. In order to solve this problem, this paper proposes a computer vision tracking algorithm combining ORB(oriented fast and rotated brief)feature points and Mean Shift algorithm. The traditional Mean Shift algorithm runs faster, but it is easy to fail while the target is blocked, which results in the tracking result and it is not satisfactory. Therefore, it is decided to use the Mean Shift algorithm which combines the ORB feature detection algorithm to achieve the target tracking. The algorithm improves the accuracy of the search window by detecting the initial position of the target and calculating the scale and rotation angle of the target in the current frame according to the feature points matched with the template. Compared with the traditional algorithm, the algorithm proposed in this paper shows a good robustness in solving the problem of obscuring the target, and it has a good performance in real-time tracking.
参考文献/References:
[1] FUKUNAGA,K,HOSTETLER L.The estimation of the gradient of a density function,with applications in pattern recognition[J].IEEE Transactions on Information Theory,1975,21(1):32-40.
[2] Comaniciu D,RAMESH V,MEER P.Real-time tracking of nonrigid object susing mean shift[C].Proc of IEEE Conf on Computer Vision and Pattern Recognition. Hilton Head Island:[s.n.],2000:142-149.
[3] Comaniciu D,MEER P.Mean shift analysis and applications[C].The Proceedings of the Seventh IEEE International Conferenceon Computer Vision.Kerkyra Greeke 1999.
[4] CHENG Yi zong.Mean shift mode seeking and clustering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1995,17(8).
[5] LOWE D G.Distinctive image features from scale-invariant keypoints[J].Interrnational Journal of Computer Vision,2004,60(2):91-110.
[6] Ke Y,Sukthankar R.PCA-SIFT:A more distinctive representation for local image descriptors Computer Vision and Pattern Recognition[C].Proceedings of the 2004 IEEE Computer Society Conference on.[s.I.]:IEEE,2004:506-513.
[7] CHEN A H,ZHU M,WANG Y H.Mean shift tracking combining SIFT[C].ICSP2008 Proceedings.Leipzig,Germany,Springer LNCS,2008,1532-1535.
[8] Dong R,Li B,Chen Q M.Multi-degree-of-freedom mean-shift tracking algorithm based on SIFT feature[J].Control and Decision,2012,27(3).
[9] 王智军,王建华.基于SIFT验证的Mean Shift跟踪运动目标新算法[J].电光与控制,2016,23(11):93-96.
[10] DI N,LI G J,CHEN CH N.Image matching technology research based on normalized DOG features[J].Journal of Electronic Measurement and Instrumentation,2014,28(6):585-560.
[11] YANG J H,HANG Y Q,XU J F,et al.Quick realization of CUDA-based registration of high-resolution digital video images[J].Chinese Journal of Scientific Instrument,2014,35(2):380-386.
[12] WEI X,FAN SH SH,GAO SH.Rapid matching and transform algorithm of images based on the BRISK[J].Electronic Measurement Technology,2015,38(3):43-47.
[13] LU M M,ZHAO F J,LI N. Feature extraction of highresolution SAR images based on bag-of-word model[J].Foreign Electronic Measurement Technology,2015,34(6):62-69.
[14] LIU S S,ZHANG H,MAO ZH,et al.Target detection method based on HRM feature extraction and SVM[J].Foreign Electronic Measurement Technology,2014,33(10):38-41.
[15] Rublee E,Rabaud V,Konolige K.ORB:an efficient alternative to SIFT or SURF[M].ICCV,2011:2564-2571.
[16] 钟华民,王伟,张慧华,等.结合ORB特征和色彩模型的视觉跟踪算法[J].模式识别与人工智能,2015,28(1):91-96.
备注/Memo
收稿日期:2017-06-06 基金项目:四川省教育厅自然科学重大培育资助项目(17CZ0007)