New version page

FengYoonJain_FuseRollPlain_ICB09

This preview shows page 1-2-3-4 out of 11 pages.

View Full Document
View Full Document

End of preview. Want to read all 11 pages?

Upload your study docs or become a GradeBuddy member to access this document.

View Full Document
Unformatted text preview:

Latent Fingerprint Matching: Fusion of Rolledand Plain FingerprintsJianjiang Feng, Soweon Yoon, and Anil K. JainDepartment of Computer Science and EngineeringMichigan State University{jfeng,yoonsowo,jain}@cse.msu.eduAbstract. Law enforcement agencies routinely collect both rolled andplain fingerprints of all the ten fingers of suspects. These two types offingerprints complement each other, since rolled fingerprints are of largersize and contain more minutiae, and plain fingerprints are less affectedby distortion and have clearer ridge structure. It is widely known in thelaw enforcement community that searching both rolled and plain fin-gerprints can improve the accuracy of latent matching, but, this doesnot appear to be a common practice in law enforcement. To our knowl-edge, only rank level fusion option is provided by the vendors. Therehas been no systematic study and comparison of different fusion tech-niques. In this paper, multiple fusion approaches at three different levels(rank, score and feature) are proposed to fuse rolled and plain finger-prints. Exp erimental results in searching 230 latents in the ELFT-EFSPublic Challenge Dataset against a database of 4,180 pairs of rolled andplain fingerprints show that most of the fusion approaches can improvethe identification performance. The greatest improvement was obtainedby boosted max fusion at the score level, which reaches a rank-1 iden-tification rate of 83.0%, compared to the rank-1 rate of 57.8% for plainand 70.4% for rolled prints.Key words: Latent fingerprint, rolled fingerprint, plain fingerprint, fu-sion, minutiae matching1 IntroductionAutomated Fingerprint Identification Systems (AFIS) have played an importantrole in forensic, law enforcement and many civilian applications. Fingerprint im-ages in AFIS can be broadly classified into three categories, namely, (i) rolled,(ii) plain/flat and (iii) latent. Figure 1 shows these three types of fingerprintimages from the same finger. Rolled fingerprint images are obtained by rollinga finger from one side to the other (“nail-to-nail”) in order to capture all theridge details of a finger. Plain fingerprints are those in which the finger is presseddown on a flat surface but not rolled. Rolled and plain impressions are obtainedeither by scanning the inked impression on paper or by directly using livescandevices. In AFIS, rolled and plain fingerprints are generally acquired in an at-tended mode and are subject to a recapture if the image quality is poor. InTo appear in ICB 2009(a) (b) (c)Fig. 1. Three types of fingerprint images. (a) Rolled fingerprint, (b) plain fingerprintand (c) latent fingerprint.contrast, latent fingerprints are lifted from surfaces of objects that are inadver-tently touched or handled by a person through a variety of means ranging fromsimply photographing the print to more complex dusting or chemical processing[1]. With small area, unclear ridge structure, complex background and strongdistortion, latent fingerprints generally have the worst image quality among thethree types of fingerprints. However, it is the matching of a latent fingerprintagainst a database of rolled/plain fingerprints that is of utmost importance inforensics and law enforcement to apprehend suspects.AFIS may work in automatic or semi-automatic mode depending on specificapplications. In civil background check application, 10 finger impressions (rolledor plain) are submitted and the AFIS automatically returns the mated subjector reports no matches found. In suspect identification application, latents aresubmitted and the system generally returns a list of top candidates, which arethen reviewed by latent experts.Fingerprint image quality has a significant impact on system accuracy. Sincethe quality of latents is not controllable, it is important to ensure that enrolledfingerprints in the database have as clear ridge structure as possible. During theenrollment, the rolled and plain fingerprints of the ten fingers of a person areobtained through fourteen impressions. These impressions can be captured byusing traditional tenprint cards or livescan devices. Traditional tenprint cards(see Fig. 2) contain the rolled impressions of the ten fingers as well as four slapimpressions: the left slap (four fingers of the left hand), the right slap (fourfingers of the right hand) and the thumb slaps (left and right thumbs). A seg-mentation algorithm is used to automatically segment the slaps into individualplain fingerprints. There are two reasons for including plain fingerprints in ten-print cards: (i) while rolled fingerprints contain larger size and larger numberof minutiae, plain fingerprints are less distorted and often have clearer ridges;and (ii) out-of-sequence rolled fingerprints can be easily detected by matchingrolled images to slap images [2]. With an aim to avoid the time-consuming anderror prone rolling process, a touchless 3D fingerprint sensor was proposed in2Fig. 2. A tenprint card from NIST SD29 consisting of fourteen fingerprint impressions[3], which reconstructs 3D fingerprint using stereo vision techniques and thencreates 2D rolled-equivalent fingerprints. While such a scanner is very desirable,the image quality provided by the current prototype of the scanner still needssignificant improvement [4].There have been some studies that compare matching accuracy using plainand rolled fingerprints. In the plain-to-rolled fingerprint matching experimentsconducted by NIST [5], the identification rate of 10-finger plain-to-rolled androlled-to-rolled is 97.5% and 97.9%, respectively. The Federal Bureau of Inves-tigation (FBI) [6] has conducted an experiment to evaluate the accuracy ofmatching latents against plain and rolled fingerprints. The hit rate of search-ing 250 latents in NIST SD27 against plain and rolled fingerprints in the FBI’sIntegrated Automated Fingerprint Identification System (IAFIS) is 38.8% and54.4%, respectively. Fusion at rank level leads to a hit rate of 61.2%. Given thisrelatively low performance of latent fingerprint matching [8, 9], there is a strongneed for the improvement in the fusion of plain and rolled fingerprints for latentfingerprint matching.Fusion of plain and rolled fingerprints is a type of multibiometric technique[10]. A multibiometric system can combine evidences from different sources, suchas multiple fingers [11], multiple sensors [12], multiple samples [13] and


Loading Unlocking...
Login

Join to view FengYoonJain_FuseRollPlain_ICB09 and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view FengYoonJain_FuseRollPlain_ICB09 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?