Scaling-robust fingerprint verification with smartphone camera in real-life scenarios

We propose a new scheme for accurate contactless fingerprint recognition captured with smartphone cameras under various real-life scenarios. The proposed scheme can be structured using three building blocks namely: (1) finger segmentation (2) pre-processing and scaling (3) minutiae extraction and comparison. The proposed finger segmentation scheme is based on Mean Shift Segmentation (MSS) algorithm followed by multiple metrics to accurately segment the finger from the background. We then propose a new scheme to perform the finger scaling to accurately extract the fingerprint region from the segmented finger. Finally, the comparison is carried out based on the minutiae features extracted from the scaled fingerprint images. Extensive experiments are carried out on our recently collected contactless fingerprint dataset consisting of 1800 samples from 25 subjects. In order to effectively evaluate the robustness of the proposed scheme, the whole dataset is constructed using three different smartphone´s namely: Nokia N8, iPhone 4 and Samsung S1. The experimental results have shown the effectiveness of the proposed scheme on various complex backgrounds with an Equal Error Rate of 3.74% noted on Samsung S1 smartphone camera.