Mahidol University Logo
Faculty of ICT, Mahidol University
 

Admissions

Printable Version

 

PRINTED THAI CHARACTER RECOGNITION USING TOPOGRAPHIC STRUCTURE

 

TITLE PRINTED THAI CHARACTER RECOGNITION USING TOPOGRAPHIC STRUCTURE.
AUTHOR JITRLADA ROJRATANAVIJITR
DEGREE MASTER OF SCIENCE PROGRAMME IN COMPUTER SCIENCE
FACULTY FACULTY OF SCIENCE
ADVISOR SUKANYA PHONGSUPHAP
CO-ADVISOR CHOMTIP PORNPANOMCHAI
 
ABSTRACT
Character recognition is an interesting topic for a number of researchers. Researches have been consistently conducted both in Thailand and foreign countries since the research in this field can be applied to various kinds of work, for instance, optical character recognition, license plate recognition and so on. Currently, car license plate recognition is increasingly becoming an interesting topic because it can be applied to such work as the systems of automatic parking as well as the entry and exit of a place, and the expressway's toll collection, etc. In Thailand, it has been found that, although there have been researches on character recognition for a long time, most of them are on optical character recognition concentrating on solving the problem of a wide variety of character fonts appearing in documents. Research on car license plate recognition is rarely found. It can be stated that it is just the beginning of a study in this field. A significant problem found in the research on car license plate recognition is a wide variety of environments affecting characters on license plates, for example, an illumination condition which is too dark or bright, a shadow appearing on characters, etc. This thesis proposed an approach to printed Thai character recognition using the topographic structure. It was aimed at analyzing printed Thai characters under various illumination conditions. In so doing, we employed two global features, clumpiness and miscibility, to select ten candidate characters. Then, the principle of a histogram projection was used for calculating local features. With these local features, the output character category was finally selected. The proposed approach was tested with images of printed Thai single characters derived from an image scanner, including images of characters on car license plates taken under different environments by a digital camera. The testing consequences revealed that the proposed approach could be effectively used to recognize printed Thai characters under several illumination conditions. It was found that an average rate of accurate recognition of printed Thai single character images derived from a scanner was 99.52% while an average rate of correct recognition of images of characters on car license plates, taken under different environments, was 97.73%.
KEYWORD THAI CHARACTER / TOPOGRAPHIC STRUCTURE

 

Go to Top

 

ICT Building, Mahidol University, 999 Phuttamonthon 4 Road, Salaya, Nakhonpathom 73170 Tel. +66 02 441-0909 Fax. +66 02 849-6099
Mahidol University Computing Center, The Faculty of ICT, Mahidol University , Rama 6 Road, Rajathevi, Bangkok 10400 Tel. +66 02 354-4333 Fax. +66 02 354-7333