In Al-Musanna College of Technology (ACT), presently 6000 students are studying in various levels and specializations. Out of total number of students, 1000 of them are eligible to register and study their course project in the pre-final year. In this research work the mix of forged and genuine online digital signatures from 750 students of ACT were collected on the basis of 75 students in one day for a period of 10 days with the digital tablet and stored as ACT dataset. The genuine signatures obtained in every day was exchanged with the other set of 75 students for a period of 10 days to prepare the types of random, skilled and unskilled forged signatures.
5.1.1 Performance of PRS
In the below Figs explains the various performance of PRS. Here, how to login the page and how to add the students detail are explained in below
The user inputs valid User Name and Password for accessing this PRS with the assistance of login page given in Fig. 3. New users must register and Sign up their details. In case the user does not remember their password, they can use Forgot Password option to get their new password in their email. The students can view the list of project proposals and their detailed status after successful login.
Figure 4 explains to help the staff or the administrator to add students’ details and group them according to their specialization. In the add students page there are female, level and specifications. Only the new student’s details are entered, the old student’s accounts can also be deleted if it is not used for longer time. If, all the details are given, then logout the given page for unauthorized person open the considered page.
Figure 5 explains admin statistics page was used by the admin to get the statistical information and manage the student project allocation and their group members based on their level of study and specialization. Here, the number of projects by which the students is registered, and how many male and female are registered in the corresponding admin statistics page.
Figure 6 represents each student group leader must fill in their project proposal details in the online form and inform their other team members to make their digital signatures and finally submit to the project approval committee. This committee then evaluates the student’s project proposal and identifies the genuine signer’s signature by employing the proposed kernel based classifiers for recognizing the online digital signature, before approving the student’s project.
Table: 1 Accuracy analysis for various classification techniques
Classification techniques
|
Training and testing
|
Online digital signature of student
|
Accuracy
|
K-ANN
|
Training
|
|
65.8%
|
Testing
|
|
62.7%
|
K-KNN
|
Training
|
|
75.6%
|
Testing
|
|
75.3%
|
K-SOM
|
Training
|
|
94.1%
|
Testing
|
|
94.3%
|
K-SVM
|
Training
|
|
97.5%
|
Testing
|
|
97.2%
|
Table 1 explains the training and testing results in the ACT dataset for online digital signature recognition. Here, it is verified by the various classifications techniques for analyzing the performance and accuracy for signatures. The data sets were divided, into training 80% and test sets 20%. The various classification techniques are K-ANN, K-KNN, K-SOM and K-SVM. Each and every technique gives the different accuracy and performance during training and testing. For training, the signature using K-ANN technique the accuracy is 65.8%, and for testing the accuracy is 62.7%. For training, the signature using K-KNN technique the accuracy is 75.6%, and for testing the accuracy is 75.3%. For training, the signature using K-SOM technique the accuracy is 94.1%, and for testing the accuracy is 94.3%. Finally, for K-SVM the accuracy is 97.5% during training and achieves 97.2% during testing. Compared to all other techniques K-SVM gives the better performance and accuracy for various digital signatures.
Performance analysis: The accuracies of these four models and its sampling were represented and studied in the form of True Positive Rate (TPR), False Positive Rate (FPR) and Equal Error Rate (EER).
True Positive Rate (TPR): It measures the proportion of actual positives that are correctly identified the genuine signatures.
TPR = Tp /(TP + FN)
False Positive Rate (FPR): It is the probability of falsely rejecting the null hypothesis for a particular test.
TPR = Fp /(FP + TN)
Equal Error Rate (EER): EER is a biometric security system algorithm used to predetermine the threshold values for its false acceptance rate and its false rejection rate. When the rates are equal, the common value is referred to as the equal error rate.