Maximum curvature:
This method will try to extract the centre lines of the veins. This is done by calculating local maximum curvatures in cross-sectional profiles of a vein image. The method is supposed to be robust against varying vein widths and non-uniform brightness.
Repeated line tracking:
This algorithm will start at several random points and will try to track a line. If a pixel has been visited by the line tracking algorithm multiple times it is likely to be a vein. A movie showing the algorithm in action can be found here:
https://vimeo.com/51757793

Literature:
- N. Miura, A. Nagasaka, and T. Miyatake "Feature extraction of finger vein patterns based on repeated line tracking and its application to personal identification" Machine Vision and Applications.
doi: 10.1007/s00138-004-0149-2
- N. Miura, A. Nagasaka, and T. Miyatake "Extraction of finger-vein patterns using maximum curvature points in image profiles." IAPR conference on machine vision applications 9 (2005).

Usage:
The file "miura_usage.m" gives an impression of how to use the feature extraction methods. Finger region localisation can be found here:
http://www.mathworks.in/matlabcentral/fileexchange/35752-finger-region-localisation

See also:
The Biometrics Security and Privacy Group at Idiap, in Switzerland have created a Python implementation of the Miura repeated line tracker and other vein biometric algorithms. See https://www.idiap.ch/software/bob/docs/bob/bob.bio.vein/stable/index.html# for details.

Comments:
Please leave a comment, if you like it, find a bug, or know (made) good improvements to the code.