PENGENALAN TANDA TANGAN MELALUI PENGOLAHAN CITRA DIGITAL DAN JARINGAN SARAF TIRUAN RADIAL BASIS FUNCTION

Abstract

Signature is one of the each individual’s uniqueness; therefore this is often used to determine the authenticity of a document or transaction. The system is being created to efficiently recognize a signature, this also determine the function of digital image processing and Radial Basis Function neural network as implemented in the actual basis. This system is made by implementing digital image processing and the usage of Radial Basis Function neural network. Signature image that will be exercised or tested has to follow few digital image processes namely grayscale, threshold, segmentation, size normalization, thinning, and region to produce values which subsequently will be processed using Radial Basis Function neural network. The usage of Radial Basis Function neural network is started by conducting exercise toward processed signature sample from each respondent. After exercise process is finished, then a signature image owned by a respondent who has been exercised can be recognized by the system. The signature image process and Radial Basis Function neural network which has been implemented in the system successfully recognizes few signature image. This research tests 100 signature images which comes from 10 respondents which individually has provided five signatures to system exercise. The accuracy degree from testing result is 88% where there are 88 signatures that can be correctly recognized while the 12 signatures are failed to be identified.