To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
SCNN: Siamese Convolutional Neural Network for Handwritten Signature Verification |
|
Author |
Maram A. Alharthi, Khloud K. Alghamdi, Samar O. Alosaimi and Manal A. Alghamdi |
| Citation |
Vol. 25 No. 5 pp. 99-102 |
|
Abstract |
Signatures have become more appealing in recent years because of its need in daily use including checks for banks, commercial transactions, attendance, and so on. Signature has been also used for identity authentication in many fields. The handwritten signature is known as a behavioral biometric trait of users that will always be different for everyone. Two persons with identical names will usually have different signatures. A signature difference from one person to another is an identification advantage for the person. Thus, it is important to have signature verification systems. Existing systems can be divided into two types; online and offline systems. The offline system uses an image of the signature with a high chance of forging the image from the original one. Focusing on that, this paper employs the Siamese convolutional neural network (SCNN) model to verify the offline signature and detect forgery. By using different kernel sizes, epoch, and learning rates, we outperform the related works with clear margins. |
|
Keywords |
Signature, Verification, CNN, Siamese network, image |
|
URL |
