An Efficient Palmprint Based Recognition System Using 1D-DCT Features

In this paper one dimensional Discrete Cosine Transform (DCT) has been used to design an efficient palmprint based recognition system. It extracts the palmprint from the hand images which are acquired using a flat bed scanner at spatial resolution of 200 dpi. It makes use of several new techniques which helps to correct the non-uniform brightness of the palmprint, to determine the global quality of palmprint image based on contrast, focus and image resolution and to extract features using difference of 1D-DCT coefficients of overlapping rectangular blocks of variable size and variable orientation. Features of two palmprints are matched using Hamming distance while nearest neighbour approach is used for classification. The system has been tested on IITK, CASIA and PolyU databases. It performs with 100% CRR and EER less than 1% for all three databases and is found to be better than the well known palmprint systems. ∗Corresponding Author Email addresses: [email protected] (Badrinath G. S. ), [email protected] (Phalguni Gupta) Preprint submitted to IDRBT Doctoral Colloquium August 25, 2011