Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Latent fingerprint enhancement for accurate minutiae detection

Abdul Wahab, Tariq Mahmood Khan, Shahzaib Iqbal, Bandar AlShammari, Bandar Alhaqbani, Imran Razzak

TL;DR

The paper tackles latent fingerprint recognition by addressing noise, distortion, and partial data that hinder traditional matching. It introduces a GAN-based Latent Fingerprint Enhancement (LFE) framework that directly optimizes minutiae information during generation, preserving local minutiae and global ridge structure through integrated minutiae locations and orientation fields. Due to the absence of paired real data, the authors synthesize training data and employ a four-block encoder-decoder generator with a seven-block CNN discriminator to produce minutiae-faithful, ridge-preserving enhancements. Experimental results on FVC2002 and CASIA datasets show improved identification performance (e.g., CMC rank-1 up to $48\%$) and better minutiae recovery compared with state-of-the-art methods like FingerGAN, highlighting the method’s potential to advance forensic latent-to-rolled fingerprint recognition. The approach offers practical impact by enabling more robust latent fingerprint identification across sensor types and real-world forensic scenarios, with implications for the reliability of automated forensic analyses.

Abstract

Identification of suspects based on partial and smudged fingerprints, commonly referred to as fingermarks or latent fingerprints, presents a significant challenge in the field of fingerprint recognition. Although fixed-length embeddings have shown effectiveness in recognising rolled and slap fingerprints, the methods for matching latent fingerprints have primarily centred around local minutiae-based embeddings, failing to fully exploit global representations for matching purposes. Consequently, enhancing latent fingerprints becomes critical to ensuring robust identification for forensic investigations. Current approaches often prioritise restoring ridge patterns, overlooking the fine-macroeconomic details crucial for accurate fingerprint recognition. To address this, we propose a novel approach that uses generative adversary networks (GANs) to redefine Latent Fingerprint Enhancement (LFE) through a structured approach to fingerprint generation. By directly optimising the minutiae information during the generation process, the model produces enhanced latent fingerprints that exhibit exceptional fidelity to ground-truth instances. This leads to a significant improvement in identification performance. Our framework integrates minutiae locations and orientation fields, ensuring the preservation of both local and structural fingerprint features. Extensive evaluations conducted on two publicly available datasets demonstrate our method's dominance over existing state-of-the-art techniques, highlighting its potential to significantly enhance latent fingerprint recognition accuracy in forensic applications.

Latent fingerprint enhancement for accurate minutiae detection

TL;DR

The paper tackles latent fingerprint recognition by addressing noise, distortion, and partial data that hinder traditional matching. It introduces a GAN-based Latent Fingerprint Enhancement (LFE) framework that directly optimizes minutiae information during generation, preserving local minutiae and global ridge structure through integrated minutiae locations and orientation fields. Due to the absence of paired real data, the authors synthesize training data and employ a four-block encoder-decoder generator with a seven-block CNN discriminator to produce minutiae-faithful, ridge-preserving enhancements. Experimental results on FVC2002 and CASIA datasets show improved identification performance (e.g., CMC rank-1 up to ) and better minutiae recovery compared with state-of-the-art methods like FingerGAN, highlighting the method’s potential to advance forensic latent-to-rolled fingerprint recognition. The approach offers practical impact by enabling more robust latent fingerprint identification across sensor types and real-world forensic scenarios, with implications for the reliability of automated forensic analyses.

Abstract

Identification of suspects based on partial and smudged fingerprints, commonly referred to as fingermarks or latent fingerprints, presents a significant challenge in the field of fingerprint recognition. Although fixed-length embeddings have shown effectiveness in recognising rolled and slap fingerprints, the methods for matching latent fingerprints have primarily centred around local minutiae-based embeddings, failing to fully exploit global representations for matching purposes. Consequently, enhancing latent fingerprints becomes critical to ensuring robust identification for forensic investigations. Current approaches often prioritise restoring ridge patterns, overlooking the fine-macroeconomic details crucial for accurate fingerprint recognition. To address this, we propose a novel approach that uses generative adversary networks (GANs) to redefine Latent Fingerprint Enhancement (LFE) through a structured approach to fingerprint generation. By directly optimising the minutiae information during the generation process, the model produces enhanced latent fingerprints that exhibit exceptional fidelity to ground-truth instances. This leads to a significant improvement in identification performance. Our framework integrates minutiae locations and orientation fields, ensuring the preservation of both local and structural fingerprint features. Extensive evaluations conducted on two publicly available datasets demonstrate our method's dominance over existing state-of-the-art techniques, highlighting its potential to significantly enhance latent fingerprint recognition accuracy in forensic applications.
Paper Structure (14 sections, 8 equations, 4 figures, 1 table)

This paper contains 14 sections, 8 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Proposed Methodology
  4. Pre-processing
  5. Generator Architecture
  6. Discriminator Architecture
  7. Experimental Results
  8. Datasets
  9. Implementation Details
  10. Results and Discussions
  11. File naming and delivery
  12. Conclusion
  13. An example appendix
  14. Example of a sub-heading within an appendix

Figures (4)

  • Figure 1: Framework for latent fingerprint enhancement and reconstruction
  • Figure 2: CMC curve presenting the identification rates of enhanced latent fingerprints of the proposed and FingerGAN models.
  • Figure 3: Comparison of the proposed method and FingerGAN on FVC2002-DB1A maio2002fvc2002 dataset: (a) Ground truth latent fingerprint; (b) FingerGAN enhanced fingerprints; (c) Minutiae extracted from FingerGAN output; (d) Our enhanced fingerprints; (e) Minutiae from our enhanced fingerprint. Ridge endings are in red, and bifurcations are in blue.
  • Figure 4: Comparison between our approach and FingerGAN on CASIA CASIA_dataset: (a) Ground truth latent fingerprint; (b) FingerGAN enhanced fingerprint; (c) Minutiae extracted from (b); (d) Enhanced fingerprint from our method; (e) Minutiae extracted from (d). Ridge endings are in red, and bifurcations are in blue.