Delta Ruled Fully Recurrent Deep Learning for Finger-Vein Verification
P.Rakkimuthu2, M.Dharmalingam2
1P.Rakkimuthu*, Research Schalar, Department of Computer Science, Bharathiar University Arts and Science College, Modakkurichi, Erode, Tamil Nadu, India,
2Dr. M. Dharmalingam, Assistant Professor, Department of Computer Science, Bharathiar University Arts and Science College, Modakkurichi, Erode, Tamil Nadu, India.
Manuscript received on November 12, 2019. | Revised Manuscript received on 21 November, 2019. | Manuscript published on December 10, 2019. | PP: 1580-1588 | Volume-9 Issue-2, December 2019. | Retrieval Number: B7303129219/2019©BEIESP | DOI: 10.35940/ijitee.B7303.129219
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Finger-vein verification is a significant problem to be resolved in image processing because it provides high security in many practical applications. Few research works have been designed in conventional works using different machine learning techniques. However, the verification accuracy of existing algorithms was not sufficient. Also, the amount of time required for verifying the input finger vein image was more. In order to overcome such limitations, Delta Ruled Fully Recurrent Deep Learning (DRFRDL) technique is proposed. The DRFRDL technique comprises of three main layers namely input, hidden, output layer for accurate finger-vein authentication. The input layer in DRFRDL Technique takes a number of finger vein images as input and then sent it to the hidden layer. The designed DRFRDL technique used numbers of hidden layers in order to deeply examine the input finger vein images. The result of the hidden layer is feeding back into the network along with the inputs in order to find outs the vein features that exist in a given image. Followed by, the extracted vein features at hidden layers are transmitted to the output layer. In DRFRDL technique, output layer applies Gaussian activation function that calculates the features matching score via determining the association between extracted vein features and the vein features that are already stored in the database. After estimating the matching score, the output layer returns the verification result. If the output layer result is 1, then vein features are matched and the user is considered as authorized person. Otherwise, vein features are not matched and the user is considered as unauthorized person. Thus, DRFRDL technique increases the authentication performance of finger-vein with higher accuracy and minimal time. The simulation of DRFRDL Technique is conducted using metrics such as verification accuracy, verification time and false positive rate with respect to a different number of finger-vein images. The simulation results depict that the DRFRDL Technique is able to improve the accuracy and also reduces the amount of time needed for finger-vein verification when compared to state-of-the-art works.
Keywords: Delta Rule, Features Matching score, Finger-vein, Gaussian Activation Function Input layer, Hidden layer, Output layer, Recurrent Behavior.
Scope of the Article: Deep Learning