Huffman Code Function and Mahalanobis Distance-base Face Recognition

Human facial appearance is affected by lots of environmental and personal factors. The human face is a very challenging pattern to recognize because of its rigid anatomy. The main problem of face recognition is large variability of the recorded images due to pose, illumination conditions, facial expression, cosmetics, different hair styles and presence of glasses amongst others. Another major issue is the ability to project facial faces into a low sub space due to the non-linear manifold nature of face, resulting in high features, affecting the automated recognition of face. Due to the aforementioned problems this research develops an improved face recognition system using Huffman Encoding method for selecting optimal features from the high dimensional face image. Recognition of faces was carried out by obtaining the Mahalanobis distance of the test image and all the training images. The experimental results obtained showed that the method employed gave comparable recognition accuracy to existing literature.