Effects of Sensors, Age, and Gender on Fingerprint Image Quality

The performance of the widely used fingerprint recognition system is heavily influenced by fingerprint quality which in turn is impacted by different factors. Many different algorithms were developed to measure fingerprint quality. This thesis analyzes the impact of different sensors, fingerprint quality algorithms and demographic factors on fingerprint image quality. Three different fingerprint quality algorithms are tested: vendor specific ones from each sensor manufacturer, NFIQ1 and NFIQ2. Our results showed that fingerprint quality decreases with age and males have better fingerprint quality than females on most sensors. The multispectral sensor has the best and stable fingerprint image quality. NFIQ2 worked well with all the tested sensors while NFIQ1 produced anomalous results on sensor 2, 3 and 9. Vendor quality scores from some sensors are either constant or not usable. These findings can help in selecting sensors for biometric systems with targeted subjects and in improving fingerprint sensor design