Raspberry pi-based finger vein recognition system using PCANet
Loading...
Date
2018
Authors
Quek, Ee Wen
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Finger Vein Recognition System (FVRS) is a biometric technology that identifies
or verifies an individual identity based on unique vein patterns. Compared with other
biometrics, it is more secure, anti-forgery and hygiene. Thus, it successfully utilized in
many authentications nowadays. The original FVRS developed only provides verification
instead of identification. For identification, the image processing involves process of
image pre-processing, feature extraction and classification. The project utilised pre processing process such as edge detection, orientation correction and Region of Interest
(ROI) extraction that have been developed previously. The main objective in this project
is to implement a suitable feature extraction technique that can maximize the FVRS
performance. A simple deep learning network, namely Principal Component Analysis
Network (PCANet) is thus proposed. It composed of three basic data processing
components, which are PCA filter, binary hashing and histograms. PCA is employed for
learning multistage filter banks. Binary hashing and block histograms are the steps for
indexing and pooling. A comparison between PCANet and PCA shows that PCANet is
outperform under limited training samples, with an increase of 21.3% than that of PCA.
Factors which impact PCANet are studied to identify the limitations of PCANet. For
classification, k-Nearest Neighbours (kNN) with Euclidean distance algorithm is
implemented. An enhancement version for kNN algorithm, k-General Nearest
Neighbours (kGNN) have been proposed at initial stage. However, performance
comparison between kNN, kGNN and SVM shows that kNN is more suitable for FVRS
implementation. The last stage for this project is to combine previous work done into an
embedded system which can be implemented in real finger vein authentication. The
program is uploaded in the Raspberry Pi by using C++ language and OpenCV image
processing library. The performance evaluation shows that the recognition rate of FVRS
achieved 92.67% . Concluded that PCANet serve as a simple but highly competitive
baseline in finger vein recognition.