An improved finger-knuckle-print recognition using fractal dimension based on Gabor wavelet

Abstract

Finger-knuckle-print (FKP) is becoming a new challenge in biometric recognition. The FKP pattern contains multiple lines with uncertain directions, which are rich distinguishing features in biometric authentication. The accuracy is an important factor that makes the FKP recognition systems practical. In order to achieve the purpose, this paper proposes FKP recognition using fractal dimension based on Gabor wavelet (FDGW) descriptor. The proposed method is composed of two procedures, feature extraction and classification. First, FKP features are extracted by Gabor wavelet. Then the improved fractal dimension (FD) which is the best way to capture fine texture features is applied to the FKP Gabor feature images. Based on PolyU FKP database, the experimental results show that the proposed method achieves the higher recognition rate when compared with the state-of-the-art methods. The highest accuracy rate of single FKP finger is 97.29% on right middle finger.