The Study Of Stroke Extraction Models For Off-line Chinese Character Recognition

Chinese character recognition plays a very important role in the automation of document analysis. For now, there are mainly two kinds of Chinese character recognition entitled on-line and off-line. The on-line Chinese character recognition system, which accepts on-line inputs, utilizes a great amount of information such as the stroke order; write pressure and other information, while the off-line Chinese character recognition, which loses information such as stroke order and write pressure, can be applied in more occasions as long as the character image can be obtained. It faces many problems: The huge size of character set and the complicated and variable character structures. Printed characters vary in font and size, while handwritten characters vary not only in font and size, but also in writer and time. The issues raised there have assumed both scientific and commercial importance.Off-line Chinese character recognition is mainly studied through two kinds of techniques, which are based on global information and local information each. The former extracts features from Chinese character as a whole, such as moment invariants, global image transformations and all kinds of projections etc. This kinds of techniques are effective to printed Chinese character recognition while are not satisfactory to handwritten Chinese characters. Relative to the fairly big variation of global shape, the local information such as stroke and radical are more consistent. At this point, the techniques based on local information are more robust. The stroke extraction is the key step of this kind of techniques.The main works of this paper are the study of existing stroke extraction...