The Study Of Printed Numeral Recognition

Digital recognition is an important part of optical character recognition, mainly refersto the use computers to automatically recognize Arabic numerals, including the printingfonts and handwriting fonts. An effective, reliable, quick and easy digital identificationsystem has a very important commercial uses, it can be used alone or as an important partof OCR system. For example, smart security system, IC card identification system.Therefore, there is a large number of research on the recognition algorithm. the characterrecognition is an important pattern recognition research, the specific and specialapplications for character recognition is still of great research value. Here we focus onoff-line printed digital recognition. It is different from other character recognition for itsstable structure, simple shape. The purpose of this paper is to use this feature to improvethe algorithm, providing a reliable and efficient character recognition system.In this paper include the following parts:(1) based on six models of digitalrecognition system (image acquisition, image preprocessing, image binarization, Imagesplitting and normalization, character feature extraction, feature classification), researchevery stages of it.(2) improve some existing features of characters, contour features of leftand right. it has a good effect, the improved method take full advantage of it and make abetter result. The existing seven-segment method has some shortcomings, so propose anew nine-segment method (3) compare and research different classification algorithms andcomplete the digital recognition system design and implementation.