Adaptive Background Estimation for Object Tracking

2 Adaptive Background Model Tracking people has received considerable attention by computer vision researchers. Interest is motivated by the broad range of potential applications such as personal identification, human-machine interaction, and automated surveillance. We previously proposed a method for extracting moving region by subtraction from background images. Background images were estimated by using UD factorized Kalman Filter. In order to use a Kalman Filter, good initial parameters are needed for estimation and were manually given in the previous work. In this paper we propose an improved method for estimation of background image and for automatic parameter initialization. We also show experimental results using real images to test the performance of our proposed method.

[1]  Mohan M. Trivedi,et al.  Moving shadow and object detection in traffic scenes , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[2]  W. Eric L. Grimson,et al.  Adaptive background mixture models for real-time tracking , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[3]  Naoya Ohta,et al.  A statistical approach to background subtraction for surveillance systems , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[4]  Larry S. Davis,et al.  Non-parametric Model for Background Subtraction , 2000, ECCV.

[5]  Bo Zhang,et al.  A neural network approach to the elimination of road shadow for outdoor mobile robot , 1997, 1997 IEEE International Conference on Intelligent Processing Systems (Cat. No.97TH8335).

[6]  Mohan M. Trivedi,et al.  Analysis and detection of shadows in video streams: a comparative evaluation , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.