الفهرس | Only 14 pages are availabe for public view |
Abstract The need for off-line signatures is crucial in our life. It is used in many applications such as checks, certificates, contracts, and historical documents, since the signature reflects each person’s characteristics. Since each corporation has a huge amount of. signatures we need a method to allow efficient recognition and verification with high rate of correctness, In this thesis, the following problem is addressed: given a set of. signature samples, recognize them then verify if they are forged or genuine in an efficient way. The existing techniques for recognition and verification are reviewed, then a recognition technique is developed based on a multi stage classifier. Its performance is compared with the available techniques., concerning the ability to recognize signatures with low error rate. New Algorithms for signature verification. are a1s’o described and tested, based on . fuzzy concepts. It is concluded from the experimental results that each of the proposed techniques has an outstanding performance on several accounts. |