A Feature Point Based Approach for Pose Variant Face Recognition

Abstract

The Pose variation challenge with respect to missing people database scenario in computerized face recognition is addressed in this study. Moreover, relationships of 2D face images with the angle variations of 0°, 45° and 90° for the same person are obtained. A feature point based approach with geometric distances of the half of face is applied. Moreover, a mathematical model and an Artificial Neural Network model are implemented using curve fitting technique to predict the face images. The face recognition accuracy is mainly tested by using face hit ratio, with Sri Lankan test subjects.