Face Recognition System Using Fisher Faces Feature Extraction

One of the challenging problems in the field of image analysis and computer vision is face recognition. Face recognition has been developed because face is considered to be one of the most important visual objects for identification. Human Face is complex and No one has a similar absolute face, even though human identical twins. Face recognition has some problems when recognizing the faces, like pose variation, illumination variation, etc.In this thesis, Face Detection and recognition system using Viola-Jones Algorithm and Fisher Faces Feature Extraction is proposed. The system consists of four steps: image processing, face detection, feature extraction and classification. Face detection using ViolaJones Algorithm is used to detect the faces. For feature extraction is used Fisher Face method. The last step is classification the face with the match face. The method to classify the face is calculating the Euclidean distance of the face. The face which has the minimum Euclidean distance is the result. In order to solve the problems, this thesis uses fisherfaces feature extraction, because Fisher Faces is one of the feature extraction methods that have good results to recognize faces. Sample database of face images is used in this research. The database name is ‘Faces94’. The database contains of the pose variation, expression and illumination variation. The database is tested in the face recognition system to get the recognition and accuracy rates of the system.The experiment of the system consist of 4 processes and one of the face recognition rates is compared with using the pre-processing methods The experiment result shows that the accuracy of the system will be better when the quantity of the training images more a lot than the testing images. The face recognition system using pre-processing methods has better recognition rate and accuracy.