Card Characters In High-speed Identification And Verification Systems Research And Design

Nowadays, Plastic card plays a more important role in people's daily life than before and its quantity needed growing rapidly. In face, the information printed on card can be much wealthier than the card itself, so, the card producer must make sure the correctness of information. Unfortunately, there could be some mistake in the printing process. Checking by human being is not so satisfied in efficiency, accuracy, and security. Therefore, an automatic recognition system for recognizing the characters in the card is necessary.This thesis discusses a high-speed card characters recognition and verification system, which includes three sub-systems: card image capture, character recognition and character verification. The system can detect the unqualified cards and pick them out. The verification system can be easily integrated in the card product line, so the system is a real-time online verification system.According to the order of the system process, this thesis describes the algorithms and theory used in the system. Besides, a detail description of the key technique in the system is given, including image binarization, character location, extraction and normalization, Optical Character Recognition (OCR), etc.The difficulty in the implement is how to improve the processing efficiency when the high accuracy should be guaranteed. Some new methods are also been presented. For example, we assume the probable area of character firstly, and then locate the exact area based on the probable area. Since the image could be blurred and the character could be adhered, we use connected components analysis in the character extraction.After that, the thesis sketched out the design and implementation of the system. We use multi-thread system to improve the efficiency.At the end of this thesis, a summary is given and the future work is also presented.