Offline Bengali handwritten character and numeral recognition using classical zone based and modified zone based feature with MQDF classifier - A comparative study.

摘要

In present scenario of automation and computerization it is essential to keep data stored in digital form which is very easy to access, edit and circulate. But it is not possible to provide a computer every place especially in rural areas so many official documents and forms are handwritten in regional languages like Bengali. To digitized this documents and forms it is essential to recognize the handwritten characters and stored them in Unicode format to computer. To fulfill this requirement we have to extract feature from every character Feature extraction describes the relevant shape information contained in a pattern, of scanned data sheet and classify them with the stored feature of the different characters. Each character has some features, which play an important role in pattern recognition to classify and recognize properly it is very essential to extract proper feature for every character. For that reason we have to choose proper feature which must contain maximum information for a character. This paper deals with two popular character feature, zone based and directional feature and those are classified using MQDF classifier. We build up the total system in Matlab.