期刊名称：International Journal of Advanced Research In Computer Science and Software Engineering
印刷版ISSN：2277-6451
电子版ISSN：2277-128X
出版年度：2013
卷号：3
期号：7
出版社：S.S. Mishra
摘要：Palm vein recognition is one of the most desirable biometric identification techniques. Several researches have been carried out on palm vein which has led to the proposition of different techniques. This research work focuses on palm vein recognition system using Hybrid Principal Component Analysis (PCA) and Self Organizing Map (SOM). The PCA-ANN experiments were considered twice when inputs to ANN were unscaled (raw scale between 0 and 255) and scaled (scale between 0 and 0.9). The performance of the system was evaluated based on different image resolutions, different training datasets, recognition time and recognition accuracy. The unscaled PCA-ANN and scaled PCA-ANN gave an optimal recognition accuracies of between (55% and 98%) and (56%-99%) respectively at a resolution of between 30*30 and 60*60 pixels level of cropping. Also further experiments were performed in determining the error rates so that the scalability of the algorithms to the task of controlling access will be investigated. The FAR and FRR were between (2.5%-12.5% for unscaled and 2.5-15% for scaled) and (2%-82% for Unscaled and 1%-81% for scaled) at 0.0001 threshold respectively. EER was 9.839% for unscaled PCA-ANN at 49.53 pixels resolution and 12.53% for the scaled PCA-ANN at 46.37 pixels resolution. This showed that EER was achieved at lower pixels resolution (46.37) for scaled PCA-ANN than the unscaled PCA-ANN (49.53) which revealed that overall system accuracy would optimally be attained by scaled PCA-ANN than the unscaled PCA-ANN
