摘要:
Banking transactions are of multiple types and checks are commonly used method for business-to-business transactions. Checks are physical document of payment transfer authenticated by the signature of the account holder. To verify a check, banks manually compare the signature with signature template of the account holder. Signatures tend to have a variation based on the mood, health, etc., of the person; no two genuine signature of same person are identical. With advancement in technology, the forging of signatures have become more sophisticated. Due to these factors, the manual verification of a signature is very challenging. Also with increase in the volume of the transactions requiring verification, it has become a herculean task to manually check and process each signature leading to the need of an automated system to identify forged signatures with speed and accuracy. In this paper, we are proposing a classification methodology to automatically detect forged signatures from the genuine signatures with low FAR (False Acceptance Rate).
展开