Study of Multi Target Tracking Algorithm Based on CRF Model

Abstract:

Moving target detection and tracking in complex background is the key technology in the field of computer vision, which has become one of the focus researches for many scholars at home and abroad. Many applications, such as robot navigation, video tracking, are closely related with the moving object detection and tracking in complex background. In this paper, we improve the traditional stochastic model and target matching algorithm, combining with the feature optical flow method, to detect and track moving target detection in complex scene, and get online modified CRF model. It provides theoretical support and guidance technology for future research.

• Periodical: Applied Mechanics and Materials (Volumes 556-562)
• Pages: 3860-3863
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.556-562.3860
• Online since: May 2014
• Authors: Chang Hui Wang*
• Keywords: Algorithm (GA), CRF Model, Object Matching, Target Tracking

References

[1] Cen Feng, Qi Feihu. Geodesic active contour tracking algorithm research - used in complex background and non rigid moving object tracking [J]. Computer research and development, 2011, 40 (2): 283-288.

[2] X. Ren, C. Fowlkes, J. Malik. Cue integration for figure/ground labeling [C]. Proc. Neural. Information Processing Systems (NIPS), 2010: 1121-1128.

[3] S. Kumar, M. Hebert. Discriminative random fields: a discriminative framework for contextual interaction in classification [C]. Proc. IEEE Int. Conf. Computer Vision, 2011: 1150-1159.

DOI: 10.1109/iccv.2003.1238478

[4] Jiang Shuhong, Wang Qin, Zhang Jianqiu, et al. Tracking algorithm based on center distance of target weighting and image feature recognition [J]. Journal of electronic, 2011, 34(7): 1175-1180.

[5] Hou Zhiqiang, Han Chongzhao. A survey of visual tracking [J]. Journal of automatic, 2012, 32(4): 603-617.

[6] KoS, LeeS, Jeon S et a1. Fast digital image stabilizer based on gray-code bit-plane matching IEEE Trans [J]. Consumer Electronics, 2011, 45(3): 598-603.

DOI: 10.1109/30.793546

[7] Zuo Junyi, Pan Quan, Liang Yan. Adaptive background modeling method based on model switching [J]. Journal of automation, 2010, 33(5): 467-473.

[8] Dai Kexue, Li Guohui, Tu Dan et al. Study on the current situation and prospect of background subtraction technology of video moving object detection [J]. Journal of Chinese image and graphics, 2012, 1(7): 919-927.

[9] Wu Si, Lin Shouxun, Zhang Yongdong. Automatic segmentation of moving objects in video sequences based on dynamic background construction [J]. Journal of computer, 2011, 28(8): 1386-1392.

[10] Chang Baobao. The maximum entropy model Natural Language Processing [D]. Computational Linguistics Institute of Peking University, 2010: 3-15.

[11] Xiang Xiaowen. Chinese named entity recognition based on conditional random field [D]. Xiamen University, 2011: 2-14.