Wire frame fitting for automatic tracking in model-based video coding

Model-based video coding requires the application of both image processing and machine vision techniques for proper fitting of the semantic model and its subsequent tracking throughout the rest of the sequence of a certain type (e.g. ‘head-and-shoulders’ or ‘head-only’). In this article a method of automatic semantic wire-frame fitting based on a reference database of facial images is presented. The method has been tested on a widely used data-base of images with very good results. It was possible to accurately retrieve the position of the facial features in all cases. The position of the facial features in initial frames can subsequently be used in automatic tracking. Results of automatic fitting are presented as a part of this contribution. Experimental results are also available on-line in the form of compressed movies from our Internet site at http://www.ee.ed.ac.uk/∼plma/.