Feature Extraction on Offline Handwritten Signature using PCA and LDA for Verification System

• Main Author: Muda, A. K.
• Format: Conference or Workshop Item
• Language: English
• Published: 2012
• Subjects: T Technology (General)
• Online Access: http://eprints.utem.edu.my/id/eprint/5671/1/Feature_Extraction_on_Offline_Handwritten_Signature_using_PCA_and_LDA_for_Verification_System.pdf; http://eprints.utem.edu.my/id/eprint/5671/

Handwritten signature verification system is one of biometric alternative solution for conventional vulnerable verification system. Biometric verification system has more secured since the difficulties in forgery and burglary. Handwritten signature is one of the biometric identification that frequently used in many applications because of its convenient to use. Nevertheless, handwritten signature is a behavioral biometric that easily to be forged by others. Each handwritten signature has its own characteristics based on behavioral, keystroke, and personal psychological. This research aimed to applyPCA and PCA+LDA to extract the offline handwritten signature and used the extracted feature in verification system. This research also compare the false accept rate (FAR) and false reject rate (FRR) to determine the better technique for feature extraction in verification system. The result showed that extracting the offline handwritten signature by using hybridizing PCA and LDA is better than PCA because PCA+LDA has lower false accuracy rate (FAR) in verification system.