Conventional touch-based sensors cannot capture the fingerprint images of all five fingers simultaneously due to their flat surfaces because the view of the thumb is not parallel to the other fingers. In addition, touch-based sensors have inherent difficulties, including variations in captured images due to partial contact, nonlinear distortion, inconsistent image quality, and latent images. These degrade recognition performance and user acceptance to using sensors. To overcome these difficulties, we propose a device that adopts a contact-free structure composed of a charge-coupled device (CCD) camera, rotating mirror equipped with a stepping motor, and a green light-emitting diode (LED) illuminator. The device does not make contact with any finger and captures all five fingerprint images simultaneously. We describe and discuss the structure of the proposed device in terms of four aspects: the quality of captured images, verification performance, compatibility with existing touch-based sensors, and ease of use. The experimental results show that the proposed device can capture all five fingerprint images with a high throughput (in 2.5 s), as required at the immigration control office of a country. Also, on average, a captured touchless image takes 57% of a whole rolled image whereas the image captured from a conventional touch-based sensor only takes 41% of a whole rolled image, and they have 63 and 40 true minutiae on average, respectively. Even though touchless images contain 13.18-deg rolling and 9.18-deg pitching distortion on average, 0% equal error rate (EER) is obtained by using five fingerprint images in verification stage.
All Science Journal Classification (ASJC) codes
- Atomic and Molecular Physics, and Optics