We have outlined a technique for improving the stability of face identification and tracking when subject to variation in pose, expression and lighting conditions. The technique makes use of the observed effect of these types of variation in order to provide a better estimate of identity, and thus provides a method of using the extra information available in a sequence to improve classification. By correctly decoupling the individual sources of variation, it is possible to develop decoupled dynamic models for each. The technique we have described allows the initial approximate decoupling to be updated during a sequence, thus avoiding the need for large numbers of training examples for each individual.
Gareth J Edwards