Fusion of geometric and texture features for side-view face recognition using svm

Abstract

Biometric recognition systems have been getting a lot of attention in both academia and the industrial sector, one of such aspects of biometrics attracting interest is side-view face recognition, the side-view of the face is known to hold unique biometric information of subjects. This study embarks on contributing to the research of side-view face biometrics by proposing the fusion of geometric and texture features of the side-view face. Local Binary Pattern (LBP) was used for the extraction of texture features and the application of Laplacian filter was used for the extraction of geometric features, both features were tested in side-view face recognition individually before fusion of the two features in order to observe and note the effect of fusing the two features has on the performance of side-view face recognition, the experiments carried out in the proposed recognition system utilized Support Vector Machine (SVM) for classification, the training of the system was done using the histograms of the texture and geometric features extracted and labelled for every individual subject in the dataset. All experiments were done on the National Cheng Kung University (NCKU) faces dataset.