Enhancement of Feature Extraction with a Robust Matching for Vascular Identification

Abstract

This thesis proposes a technique to find vein patterns from near- infrared palm and wrist images and extract features for biometric recognition. The technique analyzes the geometry of the hand to isolate the vein regions and extract features based on repeated line tracking algorithm and maximum curvature algorithm. For matching, it using robust template matching technique.
The major sections in this thesis consist of pre- processing, feature extraction and matching. The pre- processing and post- processing stages present operations involving image segmentation, mathematical morphology, etc. Basically, the features extracted are used as geometric and descriptive representation of the vein patterns. In order to extract the patterns for lineshaped image, two of the algorithms, namely, repeated line tracking algorithm and maximum curvature of points are used. The line tracking algorithm has a concept of tracking the dark line pattern at random points and repeat the process until it collects all the pixel points. In contrary, maximum curvature identifies the curvature structure in vein centers and uses it’s position to draw the pattern. The matched of palm and wrist vein images is carried out, in order to determine/ verify the person’s identity. For matching, robust template matching where, pixel- to- pixel comparison has been made. On comparison with the response time of the mentioned algorithms, the system is hence, feasible for real- time aplication.