Signature Recognition System for Student Attendance System in UTP

This paper proposes an off-line signature recognition system for student attendance
system in Universiti Teknologi PETRONAS (UTP). In current system, attendance sheet
is passed across the class and students are required to signed on the paper. Later,
lecturers will check on the paper and mark any empty column. However, lecturers
always busy and seldom have time to check each signature. Basically, the system has the
ability to imitate humans' capability of recognizing signatures. Thus, it could help
lecturers in recognizing students' signatures. The system employs artificial neural
networks for recognition and training process. This system is developed mainly using
Visual Basic 6.0 and involves four basic steps, which are image acquisition, image pre
processing, and enrolment and verification process. It has two phases, training and
recognition. Both process use artificial neural network. The system was satisfactory in
all cases where there were two different signatures to be recognized with False Rejection
Rate (FRR) for genuine signature is 4% and False Acceptance Rate (FAR) for forged
signature is 28%.