(Special Paper) 21 6, 2016 11 (JBE Vol. 21, No. 6, November 2016) http://dx.doi.org/10.5909/jbe.2016.21.6.913 ISSN 2287-9137 (Online) ISSN 1226-7953 (Print) a), a), a) Visual Object Tracking by Using Multiple Random Walkers Juhyeok Mun a), Han-Ul Kim a), and Chang-Su Kim a) (multiple random walkers). (support vector machine)...,,.,.. Abstract In this paper, we propose the visual tracking algorithm that takes advantage of multiple random walkers. We first show the tracking method based on support vector machine as [1] and suggest a method that suppresses feature vectors extracted from backgrounds while preserve features vectors from foregrounds. We also show how to discriminate between foregrounds and backgrounds. Learned by reducing influences of backgrounds, support vector machine can clearly distinguish foregrounds and backgrounds from the image whose target objects are similar to backgrounds and occluded by another object. Thus, the algorithm can track target objects well. Furthermore, we introduce a simple method improving tracking speed. Finally, experiments validate that proposed algorithm yield better performance than the state-of-the-art trackers on the widely-used benchmark dataset with high speed. Keyword : Computer Vision, Tracking, Random walkers a) (School of Electrical, Korea University) Corresponding Author : (Chang-Su Kim) E-mail: cskim@mcl.korea.ac.kr Tel: +82-2-3290-3806 ORCID: http://orcid.org/0000-0002-4276-1831 2015 () (No. NRF-2015R1A2A1A10055037) ICT (No. IITP-2016-R2720-16-0007) 2016. Manuscript received September 19, 2016; Revised October 25, 2016; Accepted October 25, 2016. Copyright 2016 Korean Institute of Broadcast and Media Engineers. All rights reserved. This is an Open-Access article distributed under the terms of the Creative Commons BY-NC-ND (http://creativecommons.org/licenses/by-nc-nd/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited and not altered.
(JBE Vol. 21, No. 6, November 2016). [1,2,3,7,8]. [9]..... (multiple random walkers) [4].,.... 2. 3, 4.. 1., [3],. 64. 24, 8. 2048. [1].. (1) [6], (2). min max,.,. 0.7. 2. 2,048.,..., 64., A.
A,. A. exp x x p b r.,.,. r. A r r p x x. [4].. (6) (7). p f Ap f r f p b Ap b r b p f r r r p x x N x k x x,.. (8) 0 1, r (9), 1. : Fig. 1. An example of patch weighs. Red depicts a high patch weight, while blue depict a low patch weight
(JBE Vol. 21, No. 6, November 2016) RWR(Random Walker with Restart). r. 0.9. x k. k x (12). (likelihood). x x p.. (6) (7). (15). p f p b c 0.7.. 1 (10). 3.,,.. 64...,,.. (16) (17).,. 20% 4 2. (a) [1,3] (b) Fig. 2. (a) Search region of previous algorithm [1,3] (b) Proposed search region
3. TB-100 : (a) (b) Fig. 3. Overall performances on the TB-100: (a) Precision (b) Success rate. coarse,.. 2(b), 1 (18). coarse, 1. 20%, 4. III. TB-100 [5] [1,2,3]. Intel i7 2.6GHz CPU, 16GB RAM PC. (precision) (success rate). 20. 3 [5] [1,2,3].. 4. 4(a) 4(b). 4(c) 4(d), 4(e) 4(f),.., 4(e).
918 방송공학회논문지 제21권 제6호, 2016년 11월 (JBE Vol. 21, No. 6, November 2016) 그림 4. TB-100 벤치마크 동영상에서의 추적 결과 예시. 붉은색 박스는 목표 객체를 표시한다. Fig. 4. Examples of tracking results on the TB-100 sequences. Red boxes contain target objects. (a) Basketball, (b) Clifbar, (c) Panda, (d) Liquor, (e) Bird1, (f) Shaking 에서 해결해야할 문제이다. IV. [2] 결론 [3] 본 논문에서는 분류기 기반의 객체 추적 기법에서 배경 이 분류기에 미치는 영향을 억제하기 위한 가중치 모델을 제안하였다. 억제할 배경 요소를 추출하기 위해 객체 박스 를 다수의 블록으로 나누어 다중 랜덤 워커 시뮬레이션을 수행하였다. 각 블록에 객체가 존재할 확률을 해당 블록의 특징 벡터에 가중치로 곱하여 배경의 영향을 억제하였다. 실험 결과는 제안하는 기법을 통해 객체 추적 벤치마크 영 상에서 추적 정확도를 크게 개선함을 보였고 기존 기법과 비교를 통해 제안 기법의 성능이 우수함을 확인하였다. 참 고 문 헌 (References) [1] S. Hare, A. Saffari, and P. H. S. Torr, Struck: Structured output tracking with kernels, in Proc. ICCV., pp. 263 270, 2011. [4] [5] [6] [7] [8] [9] [10] J. Henriques, R. Caseiro, P. Martins, and J. Batista, High-speed tracking with kernelized correlation filters, IEEE Trans. Pattern Anal. Mach. Intell., vol. 37, no. 3, pp. 583 596, Mar. 2015. H. U. Kim, D. Y. Lee, J. Y. Sim, and C. S. Kim, SOWP: Spatially Ordered and Weighted Patch Descriptor for Visual Tracking, n Proc. ICCV., pp. 3011 3019, 2015. C. W. Lee, W. D. Jang, J. Y. Sim, and C. S. Kim, "Multiple random walkers and their application to image cosegmentation," in Proc. CVPR, pp. 3837-3845, Jun. 2015. Y. Wu, J. Lim, and M.-H. Yang, Object tracking benchmark, IEEE Trans. Pattern Anal. Mach. Intell., vol. 37, no. 9, pp. 1834 1848, Sep. 2015. S. Shwartz, Shai, et al. Pegasos: Primal estimated sub-gradient solver for svm. Mathematical programming vol. 127, no. 1, pp. 3-30, 2011. B. J. Choi, B. W. Yoon, J. K. Song, and J. Park, Implementation of Pedestrian Detection and Tracking with GPU at Night-time, JBE, vol. 20, no. 3, May. 2015. J. Choi, Y. Choe, and Y. G. Kim, Histogram Equalization Based Color Space Quantization for the Enhancement of Mean-Shift Tracking Algorithm, JBE, vol. 19, no. 3, May. 2014. S. Kim, and Y. M. Ro, A Study for Improved Human Action Recognition using Multi-classifiers, JBE, vol. 19, no. 2, Mar. 2014. Tong, Hanhang, C. Faloutsos, and J. Y. Pan, Random walk with restart: fast solutions and applications. Knowletge and Information Systems, vol. 14, no. 3, 2008.
- 2016 2 : - 2016 ~ : - :, - 2014 2 : - 2016 2 : - 2016 3 ~ : - :, - 1994 2 : - 1996 2 : - 2000 8 : - 2000 8 ~ 2001 12 : USC - 2002 1 ~ 2003 7 : - 2005 8 ~ : - ORCID : http://orcid.org/0000-0002-4276-1831 - : 3D,,