Motion estimation based tracking of active contours

This paper addresses the application of active contous or snakes for robust tracking of contours in image sequences. Conventional snake approaches for tracking initialize the current frame snake with the snake obtained in the previous frame and then optimize this result considering only the current frame information. In our approach, motion estimation is embedded in the energy minimization process of the snake. This is possible introducing a new force in the definition of the energy of the snake and using a dynamic programming approach for the minimization of this energy. This allows for large displacements of the contour and for a more robust tracking. The technique has been applied for tracking facial features in a model-based coding scenario.