Multiple Fingerprint Recognition using Wavelet Packet Transform

초록

In this paper, we propose a feature extraction method for multiple fingerprint recognition. The proposed method does not require conventional preprocessing such as image enhancement, ridge segmentation and minutiae purification and extracts features directly from grey-scale fingerprint images. In the proposed method, a feature vector is obtained by computing the energies contained in each subband of the transformed image obtained by applying wavelet packets. Experimental results with a multiple-finger database show that high recognition rates are achieved using the proposed feature extraction method for multiple fingerprint recognition.