Development of Palmvein Recognition System Using Fire Fly Algorithm
  • Author(s): Famuyiwa, Kolawle Samuel. A ; Mosud Y. Olumoye ; Abisola Ayomide Olayiwola ; Dawodu, Adekunle Alani
  • Paper ID: 1703654
  • Page: 356-365
  • Published Date: 19-07-2022
  • Published In: Iconic Research And Engineering Journals
  • Publisher: IRE Journals
  • e-ISSN: 2456-8880
  • Volume/Issue: Volume 6 Issue 1 July-2022
Abstract

Palmvein technology is one of the most popular fields in pattern recognition. The most distinguishing advantage of vein features are high level of accuracy, difficult to forge and more table features. In this study, palmvein images of individuals were acquired; a Linear Discriminant Analysis and Firefly Algorithm (LDA-FA) model for feature extraction was formulated and implemented and the performance of the developed system was benchmarked with the LDA model. Five (5) palmvein images of each one hundred (100) individuals were captured using an infrared CCD sensitive camera. Linear discriminant Analysis was enhanced with Firefly Algorithm to extract sufficient features. Back Propagation Neural Network (BPNN) was used to determine the class the training and the testing image belong. 270 images were used in training the database and 230 images were used for testing the created database. The system was tested using False Positive Rate, False Negative Rate, Recognition Accuracy and Average Recognition Time. The system was tested for False Positive Rate, False Negative Rate and accuracy at threshold values of 0.25, 0.46, 0.60, 0.85. The LDA-FA achieved a false positive rate of 18.00%, 10.00%, 6.00%, 2.00%, false negative rate of 1.11%, 2.22%, 2.78%, 3.33% and accuracy of 95.22%, 96.09%, 96.52% and 96.96% at the threshold values respectively. The LDA achieved a false positive rate of 22.00%, 14.00%, 10.00%, 4.00%, false negative rate of 4.44%, 5.00%, 5.56%, 6.67% and accuracy of 91.74%, 93.04%, 93.48% and 93.91% at the threshold values respectively. The average training time generated by LDA-FA are 200.32s, 199.87s, 201.94 and 202.91 while that of LDA are 219.76s, 219.93s, 220.38s and 220.71 at the threshold values respectively. The result shows that the LDA-FA is less computationally expensive in terms of training time compared to the LDA model. The study concluded that the LDA-FA is more accurate with minimal false positive and false negative than LDA.

Keywords

Palmvein Technology, Recognition System, Firefly Algorithm.

Citations

IRE Journals:
Famuyiwa, Kolawle Samuel. A , Mosud Y. Olumoye , Abisola Ayomide Olayiwola , Dawodu, Adekunle Alani "Development of Palmvein Recognition System Using Fire Fly Algorithm" Iconic Research And Engineering Journals Volume 6 Issue 1 2022 Page 356-365

IEEE:
Famuyiwa, Kolawle Samuel. A , Mosud Y. Olumoye , Abisola Ayomide Olayiwola , Dawodu, Adekunle Alani "Development of Palmvein Recognition System Using Fire Fly Algorithm" Iconic Research And Engineering Journals, 6(1)