Influence of Gabor Wavelet Transform on Fingerprints Identification

The main goal of this study is to show the influence of preprocessing based on Gabor wavelet on the quality of identification of fingerprints. Computer simulations have been run to test the influence of scale and orientation of the wavelet on the behavior of the recognition system. By using the criterion based on the height of the correlation peak and peak to correlation energy ratio we have chosen two different scales of the Gabor wavelet for preprocessing step. The first - small scale - corresponds to the case of a low-pass filter. In the second one the scale is much bigger and the Gabor wavelet plays the role of a particular band-pass filter with angular orientation in the Fourier plane. We have tested the ability of correlation method based on the Vander Lugt and joint transform architectures. In the case of the 4f correlator the preprocessing step included convolution of the input signal with Gabor wavelet. In addition we have also tested a composite filter containing linear combination of five preprocessed images. In the case of JTC architecture wavelet filtering have been included in the joint spectrum plane.