Study On Image Processing Recognition System For Paper Currency Number

Along with th flourish and development of economy, currency in circulation becomes more and more. But counterfeit currency still is a trouble to finance in dustry that cannot be neglected. Therefore the accuracy and efficiency is very important for currency authentication.In this thesis, a method of real-time image processing system based on DSP that is used to recognizing bank paper number is introduced. The high-speed digital signal processor TMS320VC5409 is a kernel device in this system, combined with CPLD. This system not only meets the real time request but also have a better recognition accuracy.This thesis mostly studid a series of a rithmetic of image preprocessing,image segmentation and image recognition for image data, referring to advance technique, which has relation to recognition of bank paper number overseas at present. This paper implements a system for automatically inputting and recognizing bank paper number based on DSP. This system implements image data gathering and all the pre-processing and recognizing operations when a piece of bank paper gets across in high speed. The result is finally sent to the controlling and displaying part. In the paper, a detailed method of bank paper image data pre-processing and recognizing is provided with the composing of hardware and software. The pre-processing steps include bank paper image scanning, brightness equilibrating of bank paper, binary image transforming of bank paper, edge finding, declining image adjusting of bank paper, bank paper image direction judge, bank paper image smoothing and noise eliminating and bank paper image segmentation. This paper introduces a new method of searching for the edge of the bank paper, which named straight-line fiting with point average; also a rapid and exact method of character recognizing based on image recognition and process theory is presented according to the characteristics of letters and Arabic numeral. In this system the recognizing modes are few, but the high exactness and speed must be guaranteed. The work process of this system is described as the following. The first step is that turns on the power and then resets the DSP. The second step is that the BOOT program will transfer data and application in the Flash to RAM inside. In the end, DSP will start to run the arithmetic of image processing and recognizing. For it is programmed in the rule of modularization, the software of the system can be extended and transplanted easily. The application and popularization of this system will promote the normalization of currency and supervise of bank paper. At the same time, it can improve the efficiency and quality of finance service certainly.