Research Of Printed Number Automatic Recognition System Based On DSP

Coupons are widely applied to our daily activities nowadays. On these coupons, there are printed serial numbers for anti-counterfeit purpose. Each serial number conesponds to one coupon only. The process of examining these serial numbers is generally finished manually. During this process, it is very easy for workers to make mistakes, and the productivity is very low. Therefore, it is very essential to study the printed number automatic recognition system. By using this automatic detection technique of printing number, an enterprise can reduce its production cost, strengthen its competitive advantages, and produce greater social economic benefits.This project researches and analyses the RMB serial numbers. By gathering RMB number video images, and by performing noise removing, image binary, number location, character segmentation and image normalization, the printed number can be recognized automatically. The image binary algorithms mentioned in this article have a fast operation speed, and the results are remarkable. Also, they are easy to code. The printed numbers' uniform and standard features allow information to be drawn out conveniently. A template matching method and a feature matching method are used in number recognition algorithm, and a concept of degree of trust is also introduced in this paper. By using the template matching algorithm and feature identification algorithm together, the number recognition rate can be improved significantly; the computation and operation time can be reduced dramatically, and it can well suffice the requirement of real-time of the recognition system.This project selects the video developing platform as its DSP developing instrument that is based on TMS320C6416, which is a high-performance digital signal processor, and the CCS6000 software is used to write the DSP procedure. By transplanting image pre-process, binary, number location algorithm, normalization and recognition algorithms to the DSP procedure, the printed number automatic recognition system is established.The project is funded by the fund of youthful teacher with a master's degree in Xi'an University of Technology.