The impact of finger force on single and ten print fingerprint recognition
Abstract
The impact of force on acquiring a high quality fingerprint image has been systematically studied by researchers using single print optical scanners. Researchers have determined optimal capture ranges for high quality images, however, none of these optimal ranges have been tested in a direct capture methodology in terms of interaction process time, matching performance according to false accept and false reject rates using detection error trade-off curves, failures to acquire rates, or subject comfort. In addition, with the growing use of ten-print scanners by government entities such as the Department of Homeland Security as well as others, no study has attempted to answer how force impacts image quality, matching performance and subject comfort systematically across multiple force levels. The first part of this experiment examines the impact of a force based capture method based on recommendations for optimal force levels for single fingerprint capture. This method is referred to as force-capture. This force-capture method was compared to a currently existing auto-capture method in terms of image quality, minutia count and matching performance at 0.1%, 0.01%, and 0.001% FAR, process time and subject comfort during interaction. The second part of this study systematically examined the impact of applied force on ten-print scanner images at multiple force levels in terms of image quality, recognition performance at 0.1%, 0.01%, and 0.001% FAR and subject comfort during interaction. Results of this study may be able to suggest improvements to fingerprint capture algorithms, optimal force ranges for a ten-print scanner to include the four finger placements and thumb placements, and also indicate how users interact with ten-print scanners in terms of behaviors in placing their hands on the platen and the distribution of weight across the scanner.
Degree
M.S.
Advisors
Elliott, Purdue University.
Subject Area
Electrical engineering|Information science|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.