Investigation of face recognition using Gabor filter with random forest as learning framework

Face recognition on a tilted face with expression poses challenging tasks. This paper presents an investigation of face recognition based on a Gabor Filter and Oriented Gabor Phase Congruency Image with Random Forest. Gabor Filter (GF) gives the magnitude information and Oriented Gabor Phase Congruency Image (OGPCI) gives the phase information of the Gabor response. Random Forest (RF) is used as an ensemble learning framework to classify the images based on features extracted from GF and OGPCI. A face recognition technique called Gabor Classifier with Random Forest (GC-RF) is presented based on GF and OPGCI. This study used face database from Georgia Institute of Technology which contains fifteen 2D images corresponding to fifty different individuals, to evaluate the proposed system. The resolution of the images is 150∗150 pixels. The result of face recognition accuracy on the proposed system using Georgia Tech face database is 89.2%.