Research And Implementation Of Banknotes Number Recognizing Algorithm Based On DSP

With the development of economy, the very large amount of Notes Issue, themarket needed to manage the currency notes in circulation. In this paper, a RMBbanknotes number recognition algorithm is proposed according to the marketdemand. The algorithm uses the technology of digital image processing andpattern recognition to achieve real-time identification of the banknotes number,and fixed on the bill counter. Experimental results show that the proposedalgorithm has both high recognition rate and high recognition speed.Banknotes number area account for only a small part of the whole banknotesimage. In this paper digital image processing technology is used to complete thework on the collection banknotes image preprocessing. Preprocessing includesbanknotes edge detection, image tilt correction, banknotes number area decision,the grayscale image binarization, smoothing and single-character cutting. Afterpreprocessing ten images of character are get, the input data as the characterrecognition algorithm.In this paper, the different character recognition algorithm performances arecomparative and analysis. According to banknotes number identification systemfor algorithm high recognition rate and high recognition rate needs, a compositenumber identification algorithm based on template matching and eigenvaluemethod is proposed. The composite algorithm uses the number of closed-loop togroup the character set, and detection the number of closed-loop of inputcharacters. Then template matching algorithm is used in the close-loop numbercorresponding group to complete character recognition. The last, according to theresults posterior characteristics are used to do targeted correction on the results.In this paper, a fast four-connected component labeling algorithm is used to dorapid detection of the number of closed-loop, and use the expansion algorithm toguarantee the stability of the close-loop characteristic.A DSP(Digital Signal Processing) chip is used to complete the banknotesnumber identification algorithm. DSP, FPGA(Field－Programmable Gate Array)and ARM(Advanced Risc Machines) are used as the core chip in hardware,completion the entire process of the collection of the banknotes image, banknotesnumber identification and ultimately displayed on the LCD(Liquid CrystalDisplay) screen. Finally, the number identification procedures curing DSP andnumber recognition module fixed in bill counter, completing of the prototype ofthe production of number identification function bill counter. This article uses alot of banknotes to test the prototype number identification, the experimental results show that the number identification module in the recognition rate andspeed meet bill counter system requirements. The number recognition modulecan be used as independent functional modules on other systems which numberrecognition and number identification algorithm for solving the digital meterreading, ID card identification number identification problem has importantreference value.