Score based recognition of 2D images in face recognition using FDPCA

High level face recognition system is capable of matching the two dimensional face images with the images having different pose and varied facial expressions from a dataset of face models. In such process, identifying the original images involves many strategic key functions. This process involves the extraction of the related image surface, key points and depth which compared with the dataset to retrieve the original image. In this paper, we have proposed a novel technique of extracting the image components based on a Fractional Distributor Principal Component Analysis (FDPCA) method, and combining the features of cumulative integration before matching. This new technique gives improvement in accuracy when compared to the conventional method where the PCA is applied as a whole on the picture.