Performance analysis of face recognition using state of the art approaches

Face analysis plays a vital role in building human computer Interaction. The aim of this work is to explore how to exploit the temporal information in a video progression for the task of face recognition using state of art methods. In this paper, firstly, the faces are detected from the Tamil movies which are captured under different environments and locations. In the next step, the well known feature extraction algorithms like PCA, LDA, D-SIFT and LBP are applied to extract the features from the faces. Finally, in the recognition phase, the classification is done using k-NN, SVM and SRC classifiers. Extensive experimental results on Yale, AR and Movie database (MVDB) show that the D-SIFT and LBP method with SRC classifier consistently performs much better than the other methods for face recognition under severe circumstances.