Average Half Face Recognition by Elastic Bunch Graph Matching Based on Distance Measurement

Average-half-face experiments the overall accuracy of the system is better than using the original full face image. Clearly experiment shows that half face data produces higher recognition accuracy [5]. The average-half-face contain the data exactly half of the full face and thus results in storage and computational time saving. The information stored in average-half-face may be more discriminatory for face identification, especially for 3D databases [6]. Accordingly this paper is a review on the use of average-half-face and we described a system for Average-halfface recognition based on the extraction of facial fudicial points such as head, nose and ear and measuring the Euclidean distance between these features using Elastic bunch Graph matching algorithm. In this facial fudicial features on the face are head, nose and ear which are described by set of wavelet (jets) components. Image graph is a bunch graph, which is constructed between the jets. Recognition is based on the Euclidean distance measurement using bunch graph. The distance is considered as a unique factor for the specific features for each person.