Stochastic Filtering of Motion Trajectory Using a Self - organizing State Space Model

Observation noise and outliers are normally contained in motion trajectories obtained by tracking feature points in image sequences. A stochastic filtering based on state space model is used to reduce the effect of observation noise and outliers. To carry out proper state estimation, time-dependent hyper-parameters governing state estimation should be determined in accordance with motion of feature point. A self-organizing state space model is introduced to estimate hyper-parameters. In the self-organizing state space model, feature coordinates and hyper-parameters are included in state vector and they are estimated simultaneously online. Since Monte Carlo filter is used for state estimation, linear approximation for nonlinear model is not needed. Experiments are done to consider the usefulness of the proposed filter.

[1]  Keinosuke Fukunaga,et al.  Introduction to Statistical Pattern Recognition , 1972 .

[2]  N. Ichimura.  Filtering and Smoothing for Motion Trajectory of Feature Poing Using Non-Gaussian State Space Model , 2001 .

[3]  R. Martin,et al.  Robust bayesian estimation for the linear model and robustifying the Kalman filter , 1977 .

[4]  N. Gordon,et al.  Novel approach to nonlinear/non-Gaussian Bayesian state estimation , 1993 .

[5]  Jeremy MG Taylor,et al.  Robust Statistical Modeling Using the t Distribution , 1989 .

[6]  Naoyuki Ichimura,et al.  Motion segmentation using feature selection and subspace method based on shape space , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[7]  James S. Meditch,et al.  Stochastic Optimal Linear Estimation and Control , 1969 .

[8]  G. Kitagawa.  Monte Carlo Filter and Smoother for Non-Gaussian Nonlinear State Space Models , 1996 .

[9]  G. Kitagawa.  A self-organizing state-space model , 1998 .

[10]  Michael Isard,et al.  CONDENSATION—Conditional Density Propagation for Visual Tracking , 1998, International Journal of Computer Vision.

[11]  David A. Forsyth,et al.  How Does CONDENSATION Behave with a Finite Number of Samples? , 2000, ECCV.

[12]  Naoyuki Ichimura.  A robust and efficient motion segmentation based on orthogonal projection matrix of shape space , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).

[13]  Y. Bar-Shalom.  Tracking and data association , 1988 .

[14]  Pietro Perona,et al.  Monocular tracking of the human arm in 3D , 1995, Proceedings of IEEE International Conference on Computer Vision.

[15]  Takeo Kanade,et al.  A Paraperspective Factorization Method for Shape and Motion Recovery , 1994, ECCV.

[16]  Hirotugu Akaike,et al.  Likelihood and the Bayes procedure , 1980 .

[17]  Rachid Deriche,et al.  Tracking line segments , 1990, Image Vis. Comput..

[18]  Stéphane Christy,et al.  Euclidean Reconstruction: From Paraperspective to Perspective , 1996, ECCV.

[19]  Ernst D. Dickmanns,et al.  Recursive 3-D Road and Relative Ego-State Recognition , 1992, IEEE Trans. Pattern Anal. Mach. Intell..

[20]  Jun S. Liu,et al.  Sequential Monte Carlo methods for dynamic systems , 1997 .