A Tesseract-based Optical Character Recognition for a Text-to-Braille Code Conversion

Abstract

This study provided a platform that converts printed text documents into corresponding braille code that will trigger the palpable output of the braille cells. The system is composed of two main parts: the image scanner and the microcontroller-based braille platform. The image scanner captures the printed text document and performs a series of pre-processing algorithms where the processed image will be subjected to character recognition using Tesseract. It is open-source software for character recognition capable of recognizing text characters in different fonts and sizes. SimpleCV, also an open-source software for computer vision and a simpler version of an OpenCV, was utilized in pre-processing of images where binarization, filtering, edge detection, and character segmentation are performed. This will allow the microcontroller-based braille platform to interpret the printed characters from the generated braille code in ASCII format that will trigger the palpable output of the braille cells. The developed system was subjected to functionality and accuracy testing to assess its performance. Accuracy was based on the capability of the system to produce the right braille outputs that match the scanned line of text which are in Arial font. The testing was conducted utilizing the Arial font size of 12, 14, 16, 18, 20, 22, and 24. Results show that the system is capable of recognizing text with greater than 85 % accuracy starting at font size 18 with an average accuracy of 88.09 % and increases accordingly as the font size increases.