Face recognition using gradient based local feature matching

In this paper a measure that computes dissimilarity score is proposed that can be used to find out the distance between two face samples. Gradient Binary Pattern are applied over face samples to transform them into some robust representation. Later corner features are extracted and they are tracked using KL-tracking and the number of unsuccessfully tracked corners are counted between each testing and training images. Four publicly available face databases are used for system testing, viz. YALE, BERN, ORL, CALTECH.