An improved GAC model for lip contour detection

An improved model for lip contour detection is proposed in this paper, which incorporates prior shape information into geometric active contours (GAC). A new shape energy function dependent on the prior shape information is used in GAC with prior shape. And the external energy function used in this model, is depending on the information of the image gradient. Therefore, the boundary of object can be captured either by higher magnitude of the image gradient or by the minimum dissimilar between the evolutive shape and the average prior shape. Experiments on mouth images show that the accuracy improvement of the shape contour detection using GAC with prior shape is nearly 11% over GAC.