Signature recognition on bank cheques using ANN

Abstract

In India, now-a-days 95% of Indians are expected to use the bank transactions even for day-to-day requirements. Recognizing the genuine signature and finding out the fraud signature is the challenging task. Here, we have used an approach of Artificial Neural Network (ANN) to recognize the signature. In this method a signature is collected from the bank cheque by cropping the area of interest. Further it is trained and stored into the trained database. Then signatures to be tested are compared with the signatures that are stored into the test database. Area, centroid, skewness, standard deviation, mean of the signature images are the parameters used to recognize the signature. By comparing the signatures from the parameters that are derived, the system can recognize the original signature.