<![CDATA[Various attempts at securing personal information have been done in both traditional and biometric ways. And among the various ways to protect information, signatures are the most widely used in identifying and verifying personal information. Therefore, efforts should be made to be able to recognize whether the signature is genuine or false by performing detection and verification. In performing the detection process used steps consisting of preprocessing, geometric extraction features, and classification with the modified-K approach method of Nearest Neighbors as a way of verifying signatures. The preprocessing process consists of filtering, binarization, thinning, cropping, and resizing. Then extraction process geometric cirri. Before performing the extraction, zoning on the image with 3 different techniques are vertical, horizontal, and zoning 4 parts. After that is done classification for signature verification process. The result is by testing the zoning technique to determine the value of FRR and FAR of each technique. The smallest FRR value obtained is 54% and the smallest FAR value is 7%. The value is obtained by applying the vertical zoning technique. This shows that the system has a good ability in performing the verification process against fake signatures. While in the process of verification of the original signature the ability of the system is still low. So in accordance with the results obtained, to improve the ability of the system can be improved on the process of preprocessing the image.]]>
Copyrights © 2018
