Comparative Analysis of Eigenface and Learning Vector Quantization (LVQ) to Face Recognition

Abstract

Face recognition is a topic most often discussed in this era because it can be applied and developed for several needs that can be useful in daily life. Face recognition always use learning method like eigenface and learning vector quantization (LVQ). The learning process is using the face of a digital image taken from a camera in five angles for one person that will apply for dataset learning (training and learning data set), and the live image is taken from the camera for testing (testing data image). The first step image will be detected with a haar cascade and captured by a camera. It will use to processing images to get the best image and prepare to be input into the network. From the experiments with 10 testings with various parameter values, the experiment results obtained the LVQ is more accurate than eigenface to identifying faces with average accuracy 66.29% and eigenface is 56.67% with comparison 9.62%, but eigenface is faster in running time with average time 4.39 seconds and 7.38 seconds for LVQ with comparison 2.98 seconds.