Several recent research efforts in the biometrics have focused on developing personal identification using very low-resolution imaging resulting from widely deployed surveillance cameras and mobile devices. Identification of human faces using such low-resolution imaging has shown promising results and has shown its utility for range of applications (surveillance). This paper investigates contactless identification of such low resolution (∼ 50 dpi) fingerprint images acquired using webcam. The acquired images are firstly subjected to robust preprocessing steps to extract region of interest and normalize uneven illumination. We extract localized feature information and effectively incorporate this local information into matching stage. The experimental results are presented on two session database of 156 subjects acquired over a period of 11 months and achieve average rank-one identification accuracy of 93.97%. The achieved results are highly promising to invite attention for range of applications, including surveillance, and sprung new directions for further research efforts.
展开▼
