Object Tracking Method of Ochotona curzoniae Based on Guidance of Motion Information

Chen Haiyan, Zhang Aihua, Hu Shiya

Abstract

Due to the randomness and unpredictability of Ochotona curzoniae movement, the motions of Ochotona curzoniae contain smooth motions and abrupt motions in natural habitat environment. Under the circumstance of abrupt motions, the target position displacement is large between two adjacent frames. The stability and accuracy of the tracking method based on smooth motion hypothesis are difficult to guarantee. And abrupt motions are easy to cause Ochotona curzoniae tracking failure because abrupt motions violate the motion smoothness constraint. In allusion to the tracking problem of Ochotona curzoniae that smooth motions and abrupt motions coexistence, an Ochotona curzoniae tracking method based on the guidance of motion information was proposed. Considering the prior knowledge that the position displacement between two adjacent frames is smaller in smooth movement and the position displacement between two adjacent frames is larger in abrupt motion, motion information between the adjacent frames was extracted by the frame difference method at first and then the movement mode of Ochotona curzoniae was judged by motion information and appropriate sampling tracking strategy was taken to track Ochotona curzoniae. If the mode was judged to be a smooth motion mode, the Markov Chain Monte Carlo （MCMC） sampling tracking method based on the motion smoothness assumption was employed. Or else Wang-Landau Monte Carlo (WLMC) sampling tracking method used for abrupt motion tracking was adopted. The experimental results show that the proposed method can not only ensure the Ochotona curzoniae tracking performance of abrupt motions but also improve the Ochotona curzoniae tracking performance of smooth motions. The tracking success rate of proposed method was 95.49%, but the tracking success rate of WLMC method was 93.68%, the mean value and the variance of central point error in the proposed method were 13.46 and 67.89, which were 84.18% and 40.67% of those in the WLMC method, reduced by 15.82% and 59.33%, respectively.


Keywords: Ochotona curzoniae, Object tracking, Motion information, Markov chain, Monte, Carlo, Wang-Landau, Monte, Carlo

 

Download Full Text:

PDF


References


Jiang Xiaolian, Li Cuihua, Li Xiongzong. Saliency based tracking method for abrupt motion via two-stage sampling [J ]. Acta Automatica Sinica, 2014,40(6) ; 1098 - 1107. (in Chinese)

Wang Fasheng, Li Xucheng, Xiao Zhibo, et al. Hamiltonian Markov Chain Monte Carlo method for abrupt motion tracking [ J ]. Journal of Software, 2014,25(7) ; 1593 - 1605. (in Chinese)

Fukunaga Keinosuke, Hostetler Larry D. 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.

Comaniciu D, Ramesh V, Meer P. Kernel-based object tracking [ J ]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(5): 564 -557.

Li Peihua. An improved mean shift algorithm for object tracking[J]. Acta Automatica Sinica, 2007, 33(4) :347 -354.

Chen Haiyan, Zhang Aihua, Hu Shiva. The object trac king of Ochotona curzoniae based on the local texture difference operator [J]. Transactions of the Chinese Society for Agricultural Machinery, 2015,46(5) :20 -25. (in Chinese)

Li Peihua. A novel color based particle filter algorithm for object trackingf J ]. Chinese Journal of Computers,2009,32( 12) :2454-2463.

Isard M, Blake A. ICONDENSATION: unifying low-level and high-level trac king in a stochastic framework [ С] // Proceedings of European Conference on Computer Vision, 1998: 893 -908.

Khan Z, Balch T, Dellaert F. MCMC-based particle filtering for tracking a variable number of interacting targets [J ]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005, 27(11) : 1805 - 1819.

Zhao Tao, Nevatia K. Tracking multiple humans in crowded environment[ C] //Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2004,2;406 -413.

Roberts G 0, Rosenthal J S. Examples of adaptive MCMC[J]. Journal of Computational and Graphical Statistics, 2009,18(2) : 349 - 367.

К won J , Lee К M. Tracking of abrupt motion using Wang -Landau Monte Carlo estimation [C] //Proceedings of the 10 th European Conference on Computer Vision, 2008: 387 -400.

Kwon J, Lee K. Wang - Landau Monte Carlo-based tracking methods for abrupt motions [ J ]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(4) : 1011 - 1024.

Zhou Xiuzhuang, Lu Yao. Abrupt motion tracking via adaptive stochastic approximation Monte Carlo sampling[C] //Proceeding of IEEE Conference on Computer Vision and Pattern Recognition,2010:1847 - 1854.

Wang E G, Landau D P. Efficient multiple-range random walk algorithm to calculate the density of states[J]. Physical Review Letters, 2001, 86(10) s 2050 -2053.

Su Yingya, Zhao Qingjie, Zhao Liujun. Abrupt motion tracking using a visual saliency embedded particle filter [ J ] - Pattern Recognition, 2014, 47:1826 - 1834.

Li Xucheng, Wang Fasheng, Lu Mingyu. Performance evaluation of abrupt motion trac kers[ С ] // IEEE International Conference on Security, Pattern Analysis, and Cybernetics, 2014:20 -23.


Refbacks

  • There are currently no refbacks.