<![CDATA[The era of technology today,, research on biometric image is not common to do. One well researched biometric image is a face recognition (face recognition). Problems on the human face recognition is a diversity of features or shape between one another face to face. Therefore, the need for facial feature extraction and classification using a particular method so that the classification can be recognized correctly.In this study proposed feature extraction method that can overcome the problems of non-linear automatic data contained in the face image, called the Two Dimensional Linear Discriminant Analysis based on Feature Fusion Strategy (TDLDA-FFS). Not stopping on feature extraction, classification methods proposed also faces that can overcome the problems of the adaptive matrix which aims to study the benefit of weight on each - each input with the method Relevanced Generalized Learning Vector quantization (GRLVQ).This research integrates methods TDLDA-FFS and GRLVQ for face recognition. With the combination of both methods are proven to provide optimal results with a level of recognition accuracy ranged between 77.78% to 82.22% with a pilot using a databaseof facial images from the Institute of Business and Information Stikom Surabaya. While the test uses a database derived from YaleB Database achieve accuracy levels ranging from 88.89% to 94.44%.]]>
Copyrights © 2016
