Writer Identity Recognition and Confirmation Using Persian Handwritten Texts

Abstract

There are many ways to recognize the identity of individuals and authenticate them and the modern world still is looking for unique biometric features of humans. The recognition and authentication of individuals with the help of their handwriting is regarded as a research topic in recent years. It is widely used in the field of security, legal, access control to systems and financial activities. This thesis tries to examines the identification and authentication of individuals in Persian (Farsi) handwritten texts so that the identity of the author can be determined with a handwritten text, and in the authentication problem, with having two handwritten texts, it is determined that whether both manuscripts belong to a specific person or not. The proposed system for recognizing the identity of the author in this study can be divided into two main parts: one part is intended for training and the other for testing. To assess the performance of introduced characteristics, the Hidden Markov Model (HMM) is used as the classifier; thus, a model is defined for each angular characteristic. The defined angular models are connected by a specific chain network to form a comprehensive database for classification. This database is then used to determine and authenticate the author.