11-10-2012, 05:56 PM
Palmprint Authentication System Using Wavelet based Pseudo Zernike Moments Features
Palmprint Authentication System Using Wavelet.pdf (Size: 951.97 KB / Downloads: 77)
Introduction
Reliability and accuracy in personal authentication system is a dominant concern to the security world. Traditional methods using passwords, tokens, ID and smart cards will be gradually obsolesced due to their lower reliability and security. The high demand of accurate and reliable authentication solutions stirs up boundless enthusiasm from researchers and industries towards development of biometric identification technology. Nowadays, biometric technology has emerged as a cutting edge technology and is launched in a large space in the applications of physical access control, airport security, identity authentication, data security, forensics and others. Researches on the issues of fingerprint identification, iris verification, facial recognition and speech recognition have been carried out extensively in both academia and industry.
Palmprint Image Preprocessing
Normally, palmprint images captured in the image acquisition stage are gray-scale and subject to noise. Moreover, these captured images not only do contain region of interest (palmprint), but also contain region of not-interest (fingers, image background, etc.). Therefore, image preprocessing is a necessary and crucial step in palmprint verification system before processing the image. The preprocessing of our system is composed of two steps:
Wavelet Transform
Basically, wavelet transform represents image as a sum of wavelets on different resolution levels. The key upper hand of the wavelet transform is it offers high temporal localization for high frequencies while attempts good frequency resolution for low frequencies. Thus, wavelet transform is able to capture local characteristics of image and in this way we have a localized view of the image/signal’s behavior.
Palmprint Images in Wavelet Domain
In this paper, discrete wavelet transform is used to decompose the palmprint image into a multiresolution representation in order to keep the least coefficients possible without losing useful image information. The multiresolution character of the wavelet decomposition leads to superior energy compaction (high image information content) and compact constitution of decomposed image. Figure 3(a) demonstrates the decomposition process by applying two-dimensional wavelet transform of a palmprint image in level 1 and Figure 3(b) depicts two levels wavelet decomposition by applying wavelet transform on the low-frequency band sequentially.