29-03-2017, 03:03 PM
Finger vein patterns are considered as one of the most promising biometric authentication methods for your security and convenience. Most of the currently available finger vein recognition methods utilize features of a network of segmented blood vessels. As an incorrectly segmented network can degrade the accuracy of the recognition, methods based on binary patterns such as local binary pattern (LBP), local derivative pattern (LDP) and local line binary pattern (LLBP) are proposed. However, the rich directional information hidden in the finger vein pattern has not been fully exploited by existing local patterns. Inspired by the Webber Local Descriptor (WLD), this document represents a new local descriptor based on the address called Local Directional Code (LDC) and applies it to the recognition of finger veins. In LDC, the local gradient orientation information is encoded as a decimal octave number. The experimental results show that the proposed method using LDC achieves better performance than methods using LLBP.
Related video is here :