Rotation Invariant Palmprint Recognition Using Radial Harmonic Fourier Moments

Abstract

The accuracy of a Palmprint recognition system is highly depending on the extracted features from Palmprint image. The radial harmonic Fourier moments have attractive characteristics such as rotation invariant, high resistance against image noise, and less redundancy, therefore, they have the capability to provide distinct features about Palmprint image. In this work a Palmprint recognition method was proposed based on radial harmonic Fourier moments as a feature extraction descriptor to provide distinct and rotation invariant features about Palmprint images. The analysis of the results of the extensive experiments conducted over two standard Palmprint databases refers that the proposed method is rotation invariant and achieves high recognition rate even in presence of image noise. Furthermore, the proposed method outperforms other Palmprint recognition methods which utilized other orthogonal rotation invariant moments such as pseudo Zernike moments and Zernike moments.