Identification Approach Lip-Based Biometric

A robust biometric identification approach based on lip shape is presented in this chapter. Firstly, we have built an image processing step in order to detect the face of an user, and to enhance the lips area based on a color transformation. This step is ended detecting the lip on the enhanced image. A shape coding has been built to extract features of the lip shape image with original and reduced images. Those reductions have been applied with reduction scale of 3:1, 4:1 and 5:1. The shape coding points have been transformed by a Hidden Markov Model (HMM) Kernel, using a minimization of Fisher Score. Finally, a one-versus-all multiclass supervised approach based on Support Vector Machines (SVM) with RBF kernel is applied as a classifier. A database with 50 users and 10 samples per class has been built (500 images). A cross-validation strategy have been applied in our experiments, reaching success rates up to 99.6% and 99.9% for original and reduced size of lip shape, respectively; using four lip training samples per class and two lip training samples, respectively; and evaluating with six lip test samples and eight lip test samples, respectively. Those success rates were found using a lip shape of 150 shape coding points with 40 HMM states and 100 shape coding points with 40 HMM states in Hidden Markov Model, respectively, reaching with reduced lip shape image our best success, and finally, our proposal.