基於FPGA之手掌靜脈辨識系統設計與實現

Design and Implementation of Palm Vein Recognition System Based on FPGA

林裕峰(YU-FONG LIN)

指導教授 ： 朱元三

國立中正大學/工學院/電機工程學系/碩士(2016年)

摘要

由於信用卡交易或其他需加密工具之盛行，其認證方式大多都是需要記憶密碼的方式，因此會有遺忘、竊取等不便性，以目前最主流之PIN (Personal Identification Number) 碼辨識來說，常由於疏忽或遺失造成不便；因此擺脫傳統需要記憶之密碼而採用人體身上之特徵做為辨識的途徑「生物特徵辨識系統」，現今已成為一種新的趨勢；而生物辨識系統中”手掌靜脈”有不易盜取、重複性低、操作方便之特性，且「手掌靜脈」為生物辨識技術中相當可靠的辨識方法之一，因此使用影像處理技術提出一套「手掌靜脈辨識系統」。

本論文引用了中國科學院自動化研究所所提供之CASIA的手掌靜脈樣本，以及自行拍攝的手掌靜脈樣本進行分析，利用影像處理將手掌靜脈最明顯之區域進行採樣及辨識，在樣本100人/700張影像的情況下有90%的正確辨識率。本論文提出手掌靜脈辨識系統軟硬體共同設計，將部分區塊實現於FPGA上，並能快速達到身分確認。

關鍵字

生物辨識 ； 靜脈 ； 影像處理 ； 場效可程式邏輯閘陣列

並列摘要

Due to popularity of credit cards and other payment tools, the recognition system need to memory password, it still inconvenient. The PIN (Personal Identification Number) recognition usually losses because the negligence. Personal Identification in safety is more important. Thus, discarding the annoying password and establishing the uniqueness of identity, return to the “biometric recognition system” which based on the human features have been a new trend. Palm vein is not easy to steal, low repetition rate, easy to use. We use this properties to design a “palm vein recognition system”.
The palm vein recognition technology is one of the biometric identification methods. But it limited to the hardware, cost, sample people and system resources. In our thesis, we try to apply some methods of image processing to the database by CASIA and capturing pictures in real. We used the most undisturbed region of palm image which has the most obviously of palm features to recognition. We establish the palm vein image database which has 100 person and 700 total images. We have 90.38% matching rate for our database. Finally we propose a hardware/software co-design for the palm vein recognition system.

並列關鍵字

Biometric Identification ； Palm vein ； Image processing ； FPGA

參考文獻

Texture Measures with Classification Based on Kullback Discrimination of Distributions”Pattern Recognition,12th LAPR International Conference, 1994.

texture measures with classification based on featured distributions," Pattern recognition, pp. 51-59,1996.

rotation invariant texture classification with local binary patterns,"Pattern Analysis and Machine Intelligence, IEEE Transactions, pp. 971-987, 2002.

[8] Alex X. Liu, Ke Shen, Eric Torng, ” Large Scale Hamming Distance Query

[9] Yifei Wan, Anil Jain,and Lin Hong,” Fingerprint Image Enhancement: