An Efficient Approach to Enhance the Performance of Fingerprint Recognition

Abstract

Fingerprint verification systems have attracted much attention in secure organizations; however, conventional methods still suffer from unconvincing recognition rate in real noisy fingerprint images. To overcome this drawback, the combining of wavelet and contourlet coefficients are proposed as robust-to-noise and discriminative features. Contourlet transform can detect the smooth parts while wavelet coefficients are capable of representing the rough details of the images. Across Group Variance (AGV) as a feature selection method is employed to select the most discriminative ones. Next, the selected features are applied to three classifiers including Boosting Direct Linear Discriminant Analysis (BDLDA), Support Vector Machine (SVM) and Modified Nearest Neighbor (MNN). FVC2004, contains four datasets, is employed to assess the proposed method along with state-of-the-art methods in terms of Genuine Acceptance Rate (GAR) and False Acceptance Rate (FAR). The proposed method provides 98.01% GAR which outperforms the conventional methods on this dataset. Moreover, the proposed features with MNN classifier achieved 1% Equal Error Rate (EER) on all four datasets.