PERFORMANCE EVALUATION OF DISCRETE COSINE TRANSFORM (DCT) AND PRINCIPAL COMPONENTS ANALYSIS (PCA) FOR FINGERPRINT RECOGNITION SYSTEMS.

Abstract

Fingerprint recognition is the oldest biometric technique that criminal science has been using for
more than 150 years. A typical fingerprint includes several singular points called minutiae
(generally a number from 12 to 30) [1]. These specific points correspond to the places of ending,
bifurcation or crossover of ridges and valleys of the finger. Extraction of the relative positions
and orientations of these minutiae allows creating a specific signature for each user guaranteeing
a secured identification. Conflicting report on the performance of the most widely adopted feature
extraction techniques for fingerprint recognition remain an open challenge. Hence, this paper
sets to conduct the performance of DCT and PCA for fingerprint recognition. The system was
designed using MATLAB R2009b programming studio, Fingerprint images were acquired, the
acquired images was compressed by breaking it into 8 X 8 blocks of pixels, working from left to
right and top to bottom, the DCT was applied to each block to remove the redundancy between
neighboring pixels. Each block is compressed through quantization. The array of compressed
blocks that constitutes the image is stored in a drastically reduced amount of space. Training
phase and testing phase was carried out on both processes of Discrete Cosine Transform (DCT)
and Principal component Analysis (PCA) to obtain the time taken for both training and testing
of the images, with 70*70, 75*75, 80*80, 85*85 as image resolution. The results show that the
DCT exploits interpixel redundancies to render excellent decorrelation for most natural images.