Research On A Blind Reader And Its Development

With the development of information and communication technologies, the computer technology has been used to design a blind reader which can help the blind to read as normal people. Equally, conveniently and easily, the blind reader meets their urgent requirement for knowledge. Essentially, the blind reader transforms the characters into speech signals to let the blind people hear the contents instead of reading them. Costing a little, running fast and implementing easily, its core technology is the character recognition.This thesis mainly studies and implements the blind reader system. It acquires images from the texts, extracts the texts and recognizes the characters according to the image processing technology and finally generates people’s voice. For this purpose, the thesis does some reaches as follows:(1) Pre-process of Printed Chinese Character images, including graying, binarization, splitting rows columns of Chinese Character text, normalization of a single Chinese character, etc. The results of Pre-process influence the accuracy of character recognition.(2) Feature extraction of Printed Chinese characters. The thesis studies and implements the extraction of Chinese character feature, like complexity, cellular and periphery feature. According to combination of them, the Chinese characters’feature library has been established.(3) Designing the multi-layer integration of classifier and accomplished two-layer classifying of Chinese characters. The accuracy and speed of recognition is Increased.(4) Completing the structure of speech library which corresponding to the one-layer Chinese characters library and Complete the function of text reading. This thesis describes some basic arithmetic and the implementation of blind readers. Besides, it also does some simulation experiments on one-layer Chinese characters library to prove the design is feasible. The results show that the blind reader in this thesis can transform the characters in images into people’s voice. The accuracy of recognizing Chinese characters reaches99.88%and the accuracy of right speech voice reaches99.88%, too. They all meet the requirement of the design. However, the naturalness of speech voice needs to be improved.