<![CDATA[Online signatures have become one part of biometrics that is widely used to identify a person, because each person has a different signature and each signature has unique physiological characteristics. The main problem in identifying online signatures is the inconsistency in the form of a person's signature when writing a signature on a surface. Special analysis is needed regarding the main characteristics that can be used to anticipate these inconsistencies so that a person's signature can be identified correctly. This research aims to obtain features that can be used to facilitate the process of identifying signature owners. This research uses a method based on global shape features and local shape features. Global shape features are used to capture signature candidates identified as having the same or similar global form. After that, local shape features are used to measure the detailed shape similarity of each signature candidate. The trial was carried out using primary data totaling 60 signatures collected from 12 respondents. Each respondent provides five signatures which are divided into three signatures as training data and two signatures as test data. The test results of the proposed method and algorithm show an identification accuracy rate of 96.67%. This shows that the extraction method from global shape features and local shape features can be used for the development of an online biometric signature recognition system.]]>
Copyrights © 2023
