Study On Text Image Recognition Technology Based On Embedded System

With the development of the electronic technology, renascent electronic production brings great convenience to people. A new electronic applicable production- portable automatic Chinese characters reading machine is proposed, which combines advanced technology such as embedded systems, Chinese character recognition, speech synthesis, and can do reading for people. Therefor, by in-depth discussing function mechanism of reading machine, according to the development situation of nowadays electronic technology, a whole reasonable system function chart is designed, which includes hardware design and software design. For hardware, exploitation of embedded systems based on ARM7 and peripherals is realized. For software, a text image Gaussin denoising algorithm based on Higher Order Cumulant is proposed, which had solved text image pretreatment well. Using Hough Transform, declining image rectification is realized and the seperateness between text area and background area is realized using recursive cuts algorithm. At the same time, Chinese characters in text image are encoded by row and queue weighted encoding method, by which a dictionary library for recogniting Chinese character is established. It gives attention to two or more things about advantages of classic text image Chinese characters features recognition method, meanwhile fetches up backwards of features recognition algorithm in similar characters recognition, which makes the automatic reading machine designed satisfys what is needed. This paper takes up with the study of character image recognition based on embedded systems, which is combined with the TTS speech synthesis technology to realize image acquisition, image processing, character recognition and the automationy of speech synthesis.