Statistical Feature Modelling for Active Contours

A method is proposed of robust feature-detection for visual tracking. Frequently strong background clutter competes with foreground features and may succeed in pulling a tracker off target. This effect may be avoided by modelling the appearance of the foreground object (the target). The model consists of probability density functions of intensity along curve normals—a form of statistical template. The model can then be located by the use of a dynamic programming algorithm—even in the presence of substantial image distortions. Practical tests with contour tracking show marked improvement over simple feature detection techniques.

[1]  Timothy F. Cootes,et al.  Building and using flexible models incorporating grey-level information , 1993, 1993 (4th) International Conference on Computer Vision.

[2]  Richard Szeliski,et al.  Tracking with Kalman snakes , 1993 .

[3]  Arthur Gelb,et al.  Applied Optimal Estimation , 1974 .

[4]  Andrew P. Witkin,et al.  Analyzing Oriented Patterns , 1985, International Joint Conference on Artificial Intelligence.

[5]  M. Carter.  Computer graphics: Principles and practice , 1997 .

[6]  Alan L. Yuille,et al.  Deformable templates , 1993 .

[7]  Andrew Blake,et al.  Statistical Background Modelling for Tracking with a Virtual Camera , 1995, BMVC.

[8]  Rachid Deriche,et al.  Region tracking through image sequences , 1995, Proceedings of IEEE International Conference on Computer Vision.

[9]  Terry E. Weymouth,et al.  Using Dynamic Programming For Minimizing The Energy Of Active Contours In The Presence Of Hard Constraints , 1988, [1988 Proceedings] Second International Conference on Computer Vision.

[10]  Christopher J. Taylor,et al.  The use of geometric and grey-level models for industrial inspection , 1987, Pattern Recognit. Lett..

[11]  Ronald L. Rivest,et al.  Introduction to Algorithms , 1990 .