IDENTIFICATION OF SIGNATURE USING ZONING METHODS AND SUPPORT VECTOR MACHINE

Abstract

Identification of signatures is the process of identifying and defining a person’s signature. Identification of signatures including biometics that use natural human behavior. Identification of signatures can be used in security areas such as money withdrawal permits, check validation, credit card transactions and more.  During this signature identification is done manually. Difficulty in this way, if the signature to be identified is large, the examiner will experience fatigue. To simplify it needs to be developed to create a computerized signature identification system. In this research, the development of this signature identification is done using the method of zoning and Support Vector Machine (SVM) classification. Based on the tests that have been done, normal test data test resulted in recognition accuracy of 95.31%. In testing the test data with disturbance obtained accuracy of 20.31%.  While  the  testing  of artificial  signatures  generated  an  accuracy  of  70%.  In  addition  to  the  registered  signature image  pattern,  there  are  also  signature  images  that  are  not  registered  in  the  database.  The accuracy obtained in this test is 100%.