Today, as digital camera and multimedia devices are used widely, many customers needs image recognition techniques for perceiving the printed characters from business cards and photos. It is difficult to recognize the character from digital images. Th...
Today, as digital camera and multimedia devices are used widely, many customers needs image recognition techniques for perceiving the printed characters from business cards and photos. It is difficult to recognize the character from digital images. Therefore, we need to develop recognition systems which is not affected from environmental noise factor and has stable recognition ratio.
In this research, we achieved techniques which are character recognition pre-processing, Korean character recognition, English/Arabic numerals/Symbols recognition. First, character recognition pre-processing techniques manage the document to easy to perceive the character. That is binaryzation for separating between background and character, segmentation for continuous sequence of sentences, normalization for uniform character's size, classification to Korean/English/Arabic numerals/Symbols, horizontal/vertical wrong line removal, image noise cancelation, and image quality enhancement techniques. Second, Korean character recognition is composed of classifying the six type of Korean language, partitioning consonants/vowels, recognizing target consonants/vowels. Third thing is English/Arabic numerals/Symbol recognition. we used original image comparing techniques for English/Arabic numerals/Symbol recognition and type classification techniques for recognizing speed.