The Research Of Rmb Number Detection Arithmetic Based On Digital Image Processing

Banknote is the blood of finance and indispensable in the world nowadays. In recent years, the incident of bank note being hijacked happened frequently, this seriously affect the seciety safy and make the bank suffer severe lossing.So in order to reduce the bank tossing, based on the exclusive characteristic of bank note number, memorized the bank note number before put the bank note into the storeroorn,simultaneity in the currency market offer an equipment which can automationly recognize the bank note number.At a result, if the current currency bank note number is accord with the bssing ,then restrict the use and it is propitious to the detection of hijacked accident. 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.The number on the paper currency is the amount identifier of the paper currency issuance, each sheet has the number that other sheet cant has, so the number on it can identify paper currency's identity. The number is the combination of English characters and numeric characters. In this article, we design and implementation of a note number recognition by using digital image processing and pattern recognition technology.This paper 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 introduces a new method of character recognition.Character recognition includes image pre-processing, segmentation of the serial number, single character segmentation and character recognition. In this paper, after these images processing, we ultimately realize the character recognition, and simulate the results.In this paper, some important methods such as image processing, pattern recognition, computer vision and feature detection are used together to realize the number detection of paper currency image. The procedure of currency number detection mainly includes image preprocessing, number location detection, image segmentation and digital number recognition.Firstly, paper currency image pre-processing is done. The image preprocessing is necessary in the banknote identification system which includes image denoising, gray processing, image segmentation and correction, image binarization etc. In this paper, we realize the image binarization with the method of threshold.Secondly, key algorithms of paper currency recognition are studied, which includes the rough location of serial number, the extraction of serial number, single-character segmentation and character normalization. According to approximate location of the serial numbers, we can get a rough image of serial numbers. The location of serial number area can be precisely got by using the vertical and horizontal projection. Then using row and column segmentation is used to get serial number from image. The method of single-character segmentation is similar to the serial number segmentation.Finally, the character recognition algorithm is studied, this paper use a simple template matching method, after the establishment of standard character library, the character which will be recognized is compared with the templates, and find the character template which is the most similar to the tested character.At present, there are many character number reorganization methods, such as template matching method, feature analysis method and background characteristics identification method, which have advantages and disadvantages in different application conditions. Neural networks are very suitable for image recognition process, including some capabilities and features of self-organization, promotion and parallel processing. The superior characteristics of neural networks mainly represent the detection capability and recognition time of image feature. The number detection method of paper currency image based on template matching algorithm is introduced in this paper. The experimental results demonstrate that the proposed method is simple and has high number recognition efficiency.