Vein pattern recognition based on RGB images using Monte Carlo simulation and ridge tracking.

Area of Science:

• Biometrics
• Forensic Science
• Image Processing

Background:

• Identifying criminals from RGB images is challenging due to unavailable facial biometrics.
• Vein patterns offer a potential biometric solution, but are unobservable in standard RGB images.
• Traditional vein pattern acquisition uses near-infrared (NIR) technology, unsuitable for forensic RGB data.

Purpose of the Study:

• To develop a comprehensive scheme for uncovering, extracting, and matching vein patterns from RGB images for forensic applications.
• To address the limitations of existing biometric methods in forensic investigations involving RGB evidence.

Main Methods:

• Vein patterns were uncovered using Monte Carlo (MC) simulations of light transmission in a skin optical model to obtain physical parameters.
• Vein lines were extracted via ridge tracking, utilizing local gradient orientation and geometric direction, enhanced by Hessian-based Frangi filters.
• Vein pattern matching employed a modified coherent point drift (CPD) algorithm using minutiae coordinates, Gabor energy, and curvatures.

Main Results:

• The proposed vein uncovering algorithm successfully identified vein patterns from RGB images based on simulated skin optical properties.
• The vein extraction method accurately located vein lines using advanced filtering and tracking techniques.
• The modified CPD algorithm demonstrated effective vein pattern matching, outperforming existing methods in comprehensive experiments.

Conclusions:

• The developed comprehensive scheme provides a robust solution for vein pattern analysis in forensic investigations using RGB images.
• This approach overcomes the limitations of NIR-based vein identification, expanding biometric possibilities in digital forensics.
• The experimental validation confirms the superiority and effectiveness of the proposed algorithms compared to state-of-the-art methods.