Face Recognition Using Optimum-Path Forest Local Analysis

This paper deals with the automatic recognition of human faces in images. It presents a composite approach for feature extraction, evaluated using Eigenface and Fisher face methods and an Optimum-Path Forest Local Analysis classifier for face recognition methods. The main idea of this composite approach is to split the image into sub-images, and apply the face recognition step to each sub-image. The advantages of this include reduced computational effort and a better way to deal with variations in the images of the same class, such as different illumination conditions. The classifier is a variation of the Optimum-Path Forest with local analysis. Experimental results show the improvement obtained by this composite approach and several experiments with different classifiers are used to evaluate the performance of the Optimum-Path Forest Local Analysis classifier.