Genetic Algorithm with SRM SVM Classifier for Face Verification

Face verification is an important problem. The problem of designing and evaluating discriminative approaches without explicit age modelling is used. To find the gradient orientation discard magnitude information. Using hierarchical information this representation can be further improved which results in the use of gradient orientation pyramid. When combined with a structural risk minimization support vector machine with genetic algorithm, gradient orientation pyramid demonstrate excellent performance. Gradient Orientation of each color channel of human faces is robust under age progression. The feature vector which is computed as the cosines of the difference between gradient orientations at all pixels, is given as the input to the structural risk minimization support vector machine classifier. The classifier is used to divide the feature space into two classes, one for the intrapersonal pairs and the other for extrapersonal pairs. Genetic algorithm plays an important role in improving the performance of the system. The system outperformed other classifiers such as support vector machine and boosting support vector machine.