Abstract

This paper presents off-line bangle character recognition system using automatic feature extraction and XOR operation. In this system, the Bangla text is accepted as an image file which is first segmented into lines and words and then each word is segmented into characters. The pixels outside the boundary of the character are eliminated. The characters are scaled to a size equal to the database image. A XOR operation is performed between the scaled image and the database image and the error (%) is calculated. Finally, depending on the minimum error, the system recognizes the character to use in the output. The average recognition accuracy rate of the system was about 80%.