Band-Limited Phase-Only Correlation (Blpoc) Using Fpga For Finger Vein Recognition System
Nowadays, due to the high security and reliable of finger vein pattern, it had become one of the major interests in the biometric research. In the last few years, a number of finger vein recognition algorithms have been proposed. Most of the proposed methods were implemented in software-based on a g...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/46261/1/Tan%20Chee%20Wei24.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Nowadays, due to the high security and reliable of finger vein pattern, it had become one of the major interests in the biometric research. In the last few years, a number of finger vein recognition algorithms have been proposed. Most of the proposed methods were implemented in software-based on a general-purpose processor, which have limitations on the processing speed, size and power consumption. To overcome these limitations, this thesis presents an architecture for finger vein recognition system based on BLPOC matching method. The BLPOC is a phase-based matching method which have benefits of high accuracy and less affected by image shifted or brightness changed. It involves a high computation process, which is 2D-DFT, therefore, it is necessary to implement on a hardware device such as FPGA. It consists of two types of multiplexer blocks, one DFT block, one CORDIC block, seven types of memory blocks, one subtracter block, one divider block and one comparator block; and is implemented using Verilog HDL and verified using the Altera Cyclone III EP3C120F780 FPGA board. The proposed DFT block had contributed to reduce the area used by 97% of the previously proposed DFT block. A finger vein image database of 204 classes has been used to evaluate the performance of the proposed architecture. Results show that the proposed architecture can process a single matching of two finger vein images in 1.15 ms, which is about nine times faster than the softwarebased implementation, while the accuracy is similar with the software-based implementation. In conclusion, the finger vein recognition system based on BLPOC is successfully implemented on a FPGA board with better processing time as compared with the software-based implementation. |
|---|