Fingerprint Synthesis Using Deep Generative Models
(2023) In Master's Theses in Mathematical Sciences FMAM02 20231Mathematics (Faculty of Engineering)
- Abstract
- The advancements in biometric technology have amplified the need for more robust fingerprint synthesis techniques. In this thesis, we first explored the application of synthesizing normal fingerprint images in high fidelity using deep generative models (e.g., generative adversarial networks and diffusion models) and created synthetic fingerprints that retain the uniqueness and complexity of the original samples. Thereafter, by employing style transfer techniques (e.g., cycleGAN and cycleWGAN-GP), we effectively blended the global structure of one fingerprint with the local features of another fingerprint to generate a new fingerprint that is both visually realistic and distinctive, such as generating a one-to-one spoof fingerprint that is... (More)
- The advancements in biometric technology have amplified the need for more robust fingerprint synthesis techniques. In this thesis, we first explored the application of synthesizing normal fingerprint images in high fidelity using deep generative models (e.g., generative adversarial networks and diffusion models) and created synthetic fingerprints that retain the uniqueness and complexity of the original samples. Thereafter, by employing style transfer techniques (e.g., cycleGAN and cycleWGAN-GP), we effectively blended the global structure of one fingerprint with the local features of another fingerprint to generate a new fingerprint that is both visually realistic and distinctive, such as generating a one-to-one spoof fingerprint that is highly consistent with the corresponding normal fingerprint from the training set.
We employed, in the field of image generation, the state-of-the-art metrics assessing the quality of synthetic fingerprints, mainly from the perspectives of statistical analysis and subjective evaluation, and conducted a comprehensive evaluation of synthetic normal and spoof fingerprints.
Our best diffusion model achieves a promising Fréchet Inception Distance (FID) score of 15.78 and is capable of generating normal fingerprints that obtain even smaller False Acceptance Rate (FAR) than the real normal fingerprints. Our best style transfer model - cycleWGAN-GP is capable of generating spoof fingerprints in high quality from real normal fingerprints, and the distribution of these synthetic spoof fingerprints closely resembles that of real spoof fingerprints.
Our results demonstrate the potential of these methods in generating high-quality fingerprints and can be used for various applications such as security enhancement, template protection, and biometric system evaluation. (Less) - Popular Abstract
- Recent advancements in biometric technology have sparked a growing demand for more robust fingerprint synthesis techniques. This summary highlights a thesis that explores the application of deep generative models and style transfer techniques to create high-fidelity synthetic fingerprints. The study begins by utilizing deep generative models, such as generative adversarial networks (GANs) and diffusion models, to synthesize normal fingerprint images with remarkable fidelity. The goal is to retain the uniqueness and complexity of the original samples while generating visually realistic synthetic fingerprints.
Additionally, style transfer techniques, including cycleGAN and cycleWGAN-GP, are employed to blend the global structure of normal... (More) - Recent advancements in biometric technology have sparked a growing demand for more robust fingerprint synthesis techniques. This summary highlights a thesis that explores the application of deep generative models and style transfer techniques to create high-fidelity synthetic fingerprints. The study begins by utilizing deep generative models, such as generative adversarial networks (GANs) and diffusion models, to synthesize normal fingerprint images with remarkable fidelity. The goal is to retain the uniqueness and complexity of the original samples while generating visually realistic synthetic fingerprints.
Additionally, style transfer techniques, including cycleGAN and cycleWGAN-GP, are employed to blend the global structure of normal fingerprint with the local features of spoof fingerprints. This approach produces new fingerprints that possess both visual realism and distinctiveness. Notably, the study focuses on generating one-to-one spoof fingerprints that closely resemble the corresponding real spoof fingerprints from the training set while maintaining their normal fingerprint features.
To assess the quality of the synthetic fingerprints, state-of-the-art metrics in the field of image generation are employed. These metrics encompass statistical analysis and subjective evaluation, enabling a comprehensive assessment of both synthetic normal and spoof fingerprints.
For the task of normal fingerprint synthesis, the diffusion models show the best performance across all the metrics. The best style transfer model, known as cycleWGAN-GP, demonstrates the ability to generate high-quality spoof fingerprints that closely resemble real spoof fingerprints in their distribution. Out of all the metrics we experimented, FID and FAR are the most important metrics to evaluate the similarity and uniqueness of synthetic fingerprints, and the rest metrics can be used to assess the fingerprints in a wider aspect.
These findings showcase the potential of these methods in generating high-quality fingerprints for various applications. Enhanced security, template protection, and biometric system evaluation are some of the key areas where these advancements can be utilized.
In conclusion, this research presents a significant step forward in fingerprint synthesis, offering valuable contributions to the field of biometric technology. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9120105
- author
- Tang, Weizhong LU and Figueroa Llamosas, Diego André LU
- supervisor
- organization
- course
- FMAM02 20231
- year
- 2023
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Fingerprint Synthesis, Deep Generative Models, Style Transfer, Metrics
- publication/series
- Master's Theses in Mathematical Sciences
- report number
- LUTFMA-3502-2023
- ISSN
- 1404-6342
- other publication id
- 2023:E24
- language
- English
- id
- 9120105
- date added to LUP
- 2023-06-08 09:34:25
- date last changed
- 2023-07-27 17:11:54
@misc{9120105,
abstract = {{The advancements in biometric technology have amplified the need for more robust fingerprint synthesis techniques. In this thesis, we first explored the application of synthesizing normal fingerprint images in high fidelity using deep generative models (e.g., generative adversarial networks and diffusion models) and created synthetic fingerprints that retain the uniqueness and complexity of the original samples. Thereafter, by employing style transfer techniques (e.g., cycleGAN and cycleWGAN-GP), we effectively blended the global structure of one fingerprint with the local features of another fingerprint to generate a new fingerprint that is both visually realistic and distinctive, such as generating a one-to-one spoof fingerprint that is highly consistent with the corresponding normal fingerprint from the training set.
We employed, in the field of image generation, the state-of-the-art metrics assessing the quality of synthetic fingerprints, mainly from the perspectives of statistical analysis and subjective evaluation, and conducted a comprehensive evaluation of synthetic normal and spoof fingerprints.
Our best diffusion model achieves a promising Fréchet Inception Distance (FID) score of 15.78 and is capable of generating normal fingerprints that obtain even smaller False Acceptance Rate (FAR) than the real normal fingerprints. Our best style transfer model - cycleWGAN-GP is capable of generating spoof fingerprints in high quality from real normal fingerprints, and the distribution of these synthetic spoof fingerprints closely resembles that of real spoof fingerprints.
Our results demonstrate the potential of these methods in generating high-quality fingerprints and can be used for various applications such as security enhancement, template protection, and biometric system evaluation.}},
author = {{Tang, Weizhong and Figueroa Llamosas, Diego André}},
issn = {{1404-6342}},
language = {{eng}},
note = {{Student Paper}},
series = {{Master's Theses in Mathematical Sciences}},
title = {{Fingerprint Synthesis Using Deep Generative Models}},
year = {{2023}},
}