A Cross-device Matching Fingerprint Database from Multi-type Sensors Xiaofei Jia, Xin Yang, Yali Zang, Ning Zhang and Jie Tian, Fellow, IEEE Institute of Automation, Chinese Academy of Sciences {jiaxiaofei, yx, zangyali, zhangning}@fingerpass.net.cn, tian@ieee.org Abstract Databases play an important role in evaluating the performance of fingerprint identification algorithms. But which can be used to test the interoperability? That is to say, few of databases can test the performance of an algorithm on images acquired by different sensors. In order to solve the problem, we create the FingerPass cross-device matching fingerprint database which con- sists of almost 80 thousand fingerprint images from 90 subjects on nine different fingerprint sensors. We take both technology type and interaction type into consider- ation when choosing the sensors, totally different from other databases. It can test the interoperability of an algorithm at both the sensor level and the sensor type level. So we can use the FingerPass to test the perfor- mance of a cross-device matching algorithm for sensors of a specific type or different types . We apply the Ver- iFinger fingerprint recognition algorithm on it, and the experimental results indicate that the FingerPass cross- device matching database is a challenge for fingerprint algorithms. 1 Introduction Owing to reliability and stability, biometric identifi- cation is rapidly developed and widely used all over the world. Fingerprint recognition is one of the most pop- ular biometrics identification due to its high accuracy and low cost [5][10]. At the same time with the devel- opment of hardware, numerous fingerprint sensors are currently present. The consequence is that fingerprint images using for enrollment and verification may be ac- quired by different sensors. It would be in the best inter- ests of researchers in the field to develop the fingerprint recognition algorithm which has satisfied performance even for images acquired by different sensors. In this way we will have more freedom to select products. We can also use a more specialized term ”Interoperability” [8] to describe cross-device matching. But to the best of our knowledge, most of the current fingerprint recogni- tion algorithms can only operate well on a specific kind of sensors, enduring poor performance on others. With- out an algorithm that could match fingerprint images acquired by different capture sensors in enrollment and verification, all the clients attached to the same system have to be equipped with the same sensors. Now it has become an attractive challenge that how to match fin- gerprint images captured by multi-type sensors. What’ worse, it greatly hinders technological innovation in this area that there is none reliable database for cross-device matching released till now. In order to evaluate the performance of a fingerprint algorithms for the cross-device matching, we build the FingerPass database which is acquired by nine different scanners. The sensors we chose covers two differen- t technology types and two different interaction types. We can use the FingerPass to test the interoperability of an algorithm for a specific type of sensors or different types of sensors, totally different from other databases. What’s more important, our database is free to acquire for scientific research. The rest of the paper is organized as follows: Section 2 details the FingerPass database. Section 3 explains interoperability. Section 4 provides the performance of the VeriFinger fingerprint recognition algorithm on the FingerPass database. Section 5 discusses the research and points out to some potential biases of the Finger- Pass. Section 6 concludes the paper. 2 Details of the FingerPass database 2.1 Sensors of the database Nowadays, there are many ways to distinguish fin- gerprint sensors. According to technology type, these sensors can be divided into optical, capacitive, ther- mal or ultrasonic ones. And according to interaction type, they can be classified into press, sweep and non- contacted ones. As for the FingerPass, it contains nine sub-databases collected from nine biometric sensors 21st International Conference on Pattern Recognition (ICPR 2012) November 11-15, 2012. Tsukuba, Japan 978-4-9906441-1-6 ©2012 IAPR 3001