Identity Verification Based on Modeling Off line Signature Motion
Identity Verification Based on Modeling Off line Signature Motion
Arezu Labaf1 Mohammadreza Yazdchi2
1) M.SC, Department of electrical engineering, KhomeiniShahr Branch, Islamic Azad University, Khomeinishahr/Isfahan, Iran- Email:
2) Assistant Professor, Department of Biomedical Engineering, Faculty of Engineering, University of Isfahan, Isfahan, Iran-Email:
IOI Code
:
XYCZ-AAKDC
Publication :
International Conference on Engineering and Applied Science(iceasconf.com)
Abstract :
Sign exclusive features to human behavior allows it to study many different purposes, such as recognition of dynamic behavior, control and learning, detection of abnormalities of the nervous, muscular and mental , The development of psychological tests and etc should be considered. Signature moves of those gestures that are using it, are distinct individuals. Because a lot of unique traits and individual differences in the movements of the hands and fingers are lodged subtle and the cause of modeling the movement of the control system will be clear. Modeling allows you to understand, predict, and correct inference system behavior studied in laboratory provides off-limits. Among the important features and characteristics of the individual, is signature that humans use it for recognition and validation of their qualifications in correspondence with others. It is also vital in terms of legal and It is also vital legal terms and in recent years, recognition, validation and offline signature verification great importance and has attracted a lot of attention. In this study, the first to be processed and identify offline signatures based on modeling of single-layer neural networks because of the ease of implementation and proper functioning as a signature to identify the characteristics of static will be discussed. To practice identifying, each feature with the help of neural networks compared with optimum points of reference and offline signature characteristics are evaluated. In order to classify the samples will be used to support vector machine. Classification error for samples obtained 40 signatures for each user is zero.
Keywords : offline sinature neural network identification classification extraction of features