Novel methodology for Kannada Braille to speech translation using image processing on FPGA

With the introduction and popularization of text to speech convertors, a huge drop in literacy rates is being seen amongst the visually impaired. Also, since Braille is not well known to the masses, communication by the visually impaired with the outside world becomes an arduous task. A lot of research is being carried out in conversion of English text to Braille but not many concentrate on the alternative i.e. conversion of Braille to regional languages. In order to address this issue, in this paper we introduce a novel methodology to convert Braille characters representing the Kannada Language (a popular language of southern part of India), captured by a camera, into Kannada text or speech. An automated thresholding algorithm for segmentation of the Braille dots along with a novel algorithm for identification of the characters has been explained. All algorithms were designed and developed for a Xilinx Spartan 3E FPGA and were executed in real time. An accuracy of over 94% was achieved in Braille segmentation and detection. The algorithm for identification of the Kannada Braille character was found to be four times faster than many existing methodologies, on the FPGA.