The mean shift algorithm and the unified framework

This paper considers two classes of algorithms for the representation of data points using centroids: the unified framework and the mean shift algorithm. The relationship between both approaches is presented showing that the mean shift algorithm fits within the unified framework being equivalent to snake with Cohen potential. However it does not use competitive learning as the other methods considered in the unified framework. The advantages of both types of techniques are exemplified through examples.

[1]  Dorin Comaniciu,et al.  Mean Shift: A Robust Approach Toward Feature Space Analysis , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Demetri Terzopoulos,et al.  Snakes: Active contour models , 2004, International Journal of Computer Vision.

[3]  Jorge S. Marques,et al.  A constrained clustering algorithm for shape analysis with multiple features , 2000, Proceedings 15th International Conference on Pattern Recognition. ICPR-2000.

[4]  Richard Durbin,et al.  An analogue approach to the travelling salesman problem using an elastic net method , 1987, Nature.

[5]  Jorge S. Marques,et al.  A class of constrained clustering algorithms for object boundary extraction , 1996, IEEE Trans. Image Process..

[6]  David G. Stork,et al.  Pattern Classification , 1973 .

[7]  Dorin Comaniciu,et al.  Kernel-Based Object Tracking , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[8]  Laurent D. Cohen,et al.  Finite-Element Methods for Active Contour Models and Balloons for 2-D and 3-D Images , 1993, IEEE Trans. Pattern Anal. Mach. Intell..