University of Twente Student Theses

Login

Generating Realistic Ghost Fingerprints by Combining Real Fingerprint Images

Huiden, J.E. (2019) Generating Realistic Ghost Fingerprints by Combining Real Fingerprint Images.

[img] PDF
2MB
Abstract:Fingerprints are widely used biometrics in security systems and law enforcement. It is essential that possible distortions in the measurement of fingerprints are detected before the images are saved or used, to prevent incorrect matches. One possible distortion in fingerprints is called the ghost fingerprint. To train and test ghost fingerprint detection algorithms, a large amount of data is needed. Obtaining this data is not always possible due to cost, time and privacy issues. This paper proposes a method to create realistic ghost fingerprints by combining real fingerprint images which can be used as data for ghost fingerprint detection algorithms. This method uses masks to determine the region of overlap and threshold binarisation to replicate the overlapping lines as seen in real ghost fingerprint images. To create variation in the created fingerprints different transformation such as scaling, rotation, position, and changing line thickness are, within a range, randomly applied. The result is a system that is able to generate a large variety of realistic looking ghost fingerprints. Three databases are created with different classes separated by the percentage of ghost presence, the intensity of the ghost, and a combination of both. These databases are fed into two ghost fingerprint detection algorithms; one based on the local binary pattern, and one on frequency estimation. Comparing the results with real ghost fingerprints shows that the accuracy of the detection of the artificial ghost fingerprints is lower than the detection of real ghost fingerprints. The proposed separation parameters for the classes have effect on the accuracy of detection, however not as much compared to the different classes of the real ghost fingerprints.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:50 technical science in general, 54 computer science, 58 process technology
Programme:Electrical Engineering BSc (56953)
Link to this item:https://purl.utwente.nl/essays/78998
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page