Gabor Features for Single Sample Face Recognition on Multicolor Space Domain

Face recognition is growingly becoming a very remarkable field in machine learning and artificial intelligence. In this paper, we introduce a modified scheme for face recognition based on the hybrid color model along with the Gabor Feature Extraction and Principal Component Analysis (PCA). Our algorithm is tested on two face databases namely, 'The MUCT Database' and 'The FEI Database' for recognition accuracy. In addition to the various issues posed by the database such as variations in expression, illumination, ethnicity etc. we have also considered tackling and alleviating the problem of small sample size by taking only one image per person for the training set. Recognition accuracies up to 100% are obtained for MUCT database and a maximum accuracy of 87.43% is obtained for FEI database. Through these results, it is justifiable that our methodology has performed better than many of the existing algorithms specific to the databases that are considered in our experiment.