BengalI Optical Character Recognition using self organizing map

摘要

Being the 5^th Position and sweetest language in the world declared by the UNESCO Bengali is the national language in Bangladesh and one of the major languages in India. Lot of researches has been done to recognize Bengali, English and other major languages using Optical Character Recognition (OCR). To recognize Bengali character from text images and convert into editable text, Self Organizing Map (SOM) - kind of neural network has been used. To collect the character, documents are scanned, which is preprocessed with the Image to Binary Conversion Algorithm. In the binary image, character area is represented by 0 (zero) and rest of the image area is represented with 1 (one). After detecting and correcting the skew and noise, the binary image is processed and grouped, which can be mapped and recognized by SOM. Considering efficiency and fastness, character grouping process has been introduced.