Tracking Multiple Persons from Multiple Camera Images

This chapter discusses a human tracking method using multiple non-synchronous camera observations. In vision-based human tracking, self-occlusions and human-human occlusions are significant problems. Employing multiple viewpoints reduces these problems. Furthermore, the use of the non-synchronous observation approach eliminates the scalability problem inherent in synchronous systems. In the system described in this chapter, each camera independently observes a scene and thus does not require any special synchronization mechanism. The multiple observations are integrated with a Kalman-filter-based algorithm. With its non-synchronous approach, the system can obtain dense observations for the temporal axis, and the total performance is not affected by increasing the number of cameras. We developed the experimental system to accommodate five or more cameras. The system can track human positions in both single-person and multiple-person situations. Experimental results show the effectiveness of the non-synchronous multiple-camera system.

[1]  Akira Utsumi,et al.  Hand image segmentation using sequential-image-based hierarchical adaptation , 1997, Proceedings of International Conference on Image Processing.

[2]  R. E. Kalman,et al.  A New Approach to Linear Filtering and Prediction Problems , 2002 .

[3]  Tsutomu Miyasato,et al.  Hand gesture recognition system using multiple cameras , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[4]  Jakub Segen,et al.  A camera-based system for tracking people in real time , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[5]  Alex Pentland,et al.  Pfinder: real-time tracking of the human body , 1996, Other Conferences.

[6]  J. O'Rourke,et al.  Model-based image analysis of human motion using constraint propagation , 1980, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[7]  Akira Utsumi,et al.  Vision-based Human Tracking Using A Top-view Approach With Non-synchronous Multiple Observations , 2001 .

[8]  Larry S. Davis,et al.  3-D model-based tracking of humans in action: a multi-view approach , 1996, Proceedings CVPR IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[9]  K. Rohr.  Towards model-based recognition of human movements in image sequences , 1994 .

[10]  Akira Utsumi,et al.  Multiple-camera-based human tracking using non-synchronous observations , 2000 .

[11]  Masahiko Yachida,et al.  Multiple-view-based tracking of multiple humans , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[12]  Jake K. Aggarwal,et al.  Tracking human motion using multiple cameras , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[13]  Jun-Ho Oh,et al.  Recursive resolving algorithm for multiple stereo and motion matches , 1997, Image Vis. Comput..

[14]  Larry H. Matthies,et al.  Error modeling in stereo navigation , 1986, IEEE J. Robotics Autom..

[15]  Alex Pentland,et al.  Real-time self-calibrating stereo person tracking using 3-D shape estimation from blob features , 1996, Proceedings of 13th International Conference on Pattern Recognition.

[16]  Masanobu Yamamoto,et al.  Human motion analysis based on a robot arm model , 1991, Proceedings. 1991 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[17]  Atsushi Nakazawa,et al.  Human tracking using distributed vision systems , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[18]  Trevor Darrell,et al.  Evolving visual routines , 1994 .