Abstract:
Biometrics based user authentication is rather popular in mobile devices using face and fingerprints as the primary modalities. Fingerphoto (an image of a person’s finger captured using an inbuilt smartphone camera) based user authentication is an attractive and cost-e↵ective alternative. Existing research focuses on constrained or the semi-constrained environment; whereas, challenges such as user cooperation, number of fingers, background, orientation, and deformation are important to address before finger photo authentication becomes usable. This report presents (1) the first publicly available unconstrained finger photo database, termed as UNconstrained FIngerphoTo (UNFIT) database [7], which contains finger photo images acquired in unconstrained environments, (2) baseline results on the collated database using the pipeline proposed in [39], (3) deep learning-based segmentation results, (4) no-reference image quality assessment [22, 29–32, 37] and deep blind image inpainting [24], and (5) CompCode [16] and ResNet50 [12] representation based matching approaches. We further assert that the proposed UNFIT database can encourage research in this important domain.
