Braille Character Recognition Based on Neural Networks

Abstract

Braille is the most popular system used for interaction between visually-impaired and sighted people using tactile means. Optical Braille character recognition (OBCR) includes two main steps: Braille cells’ recognition (image acquisition, preprocessing, Braille dots’ recognition, Braille cells’ recognition and segmentation) and Braille cells’ transcription to corresponding natural language characters. System example has been created using image processing methods and artificial neural networks approach. These methods allow to achieve high speed and recognition accuracy level. System can adapt to factors like quality of input patterns and differences between them dynamically. In this paper, artificial neural network is developed to identify letter's images of Cyrillic alphabet in Braille representation system. Network will be trained and tested for identifying of scanned Cyrillic letters in Braille. Some of the letters are noised with some type of noise to simulate the real-world environment.