摘要: We introduce a frequency decomposition based background generation and subtraction method that explicitly harnesses the scene dynamics to improve segmentation. This allows us correctly interpret scenes would confound appearance-based algorithms by having high-variance in presence of low-contrast targets, specifically when pixels are well modeled as cyclostationary random processes. In other words, we can distinguish near-periodic temporal patterns induced real-world physics: motion plants driven wind, action waves on beach, appearance rotating objects. To capture cyclostionary behavior each pixel, compute coefficients variation pixel intensity moving windows. maintain model is composed coefficients, compare with current set obtain distance map. eliminate trail effect, fuse maps. WIAMIS 2005
参考文章(6)
Ahmed Elgammal, David Harwood, Larry Davis, Non-parametric Model for Background Subtraction Lecture Notes in Computer Science. pp. 751- 767 ,(2000) , 10.1007/3-540-45053-X_48
C. Stauffer, W.E.L. Grimson, Adaptive background mixture models for real-time tracking computer vision and pattern recognition. ,vol. 2, pp. 246- 252 ,(1999) , 10.1109/CVPR.1999.784637
Fatih Porikli, Oncel Tuzel, None, Human Body Tracking by Adaptive Background Models and Mean-Shift Analysis ,(2003)
A. Mittal, N. Paragios, Motion-based background subtraction using adaptive kernel density estimation computer vision and pattern recognition. ,vol. 2, pp. 302- 309 ,(2004) , 10.1109/CVPR.2004.1315179
C.R. Wren, A. Azarbayejani, T. Darrell, A.P. Pentland, Pfinder: real-time tracking of the human body IEEE Transactions on Pattern Analysis and Machine Intelligence. ,vol. 19, pp. 780- 785 ,(1997) , 10.1109/34.598236
K. Toyama, J. Krumm, B. Brumitt, B. Meyers, Wallflower: principles and practice of background maintenance international conference on computer vision. ,vol. 1, pp. 255- 261 ,(1999) , 10.1109/ICCV.1999.791228