An on-line hand-written character recognition system dedicated to blind people

By using a speech synthesizer, a complete dedicated personal digital assistant system can be developed for blind people. This approach is only valid for late blind people who have learned to write. However, the writing quality of blind people is very poor, mainly due to the lack of visual feedback from what they actually draw. This paper presents an adaptive pre-processing method to restore the global shape of the written words. This is obtained by adjusting the relative vertical position of the different strokes that make up a word. The pre-processing algorithm has been successfully implemented (on a Newton 130 Apple) and experimented on a dedicated database created by blind people. We observed that all strokes are fairly presented for the recognition process of the Newton 130 Apple.