Recognition Technology Correlational Research In Print Uighur Recognition System

With the development of information technology in human life, how to store anddeal with all sorts of documentation efficiently and quickly has become a hot issue incomputer science field. Nowadays, a lot of researchers have achieved great success inprinted character recognition research, and the corresponding products in differentlanguages has appeared in people’s daily life. However, because of the complicated ofthe unique structure of Uighur itself, there are a few researches on Uighur Printedcharacter recognition, the corresponding techniques are immature. The deep studiedon Uighur printed character recognition technology have a very importantsignificance on the human history culture.The paper launches the research from image preprocessing, Uighur characterfeature theory and BP neural network theory. The detail works that were done in thispaper are mainly the following aspects:(1) Discusses the background and significance of the subject. In the process ofimage preprocessing, the experiment achieved good results which using binarizationalgorithm and skew correction algorithm and propose removal discrete noisealgorithm.(2) In the character feature: firstly, the initial classification of Uighur characterwhich using itself writing characteristics and structural characteristics, the all Uighuralphabet is divided into16classifications; Secondly, mixed feature vector that form ofrough grid feature and projection feature as the input feature of BP neural network,Combining the structure feature and statistical feature of character, provides favorableconditions for the recognition.(3) In the design of BP neural network character classifier, fully absorb theresults of previous studies, the complete study from network structure, parameterdesign, neural network training and network identification. The design project which optimized BP neural network performance was finished,and verify the effectivenessof the optimization project in the final experiment.This paper shows that Uighur character recognition system based on the structurefeature and statistical feature of character not only able to effectively recognizecharacters in the same font of training samples, but can also has certain capacity ofanti-interference and anti-deformation. This design program is a reference to UighurPrinted character recognition systems.