Note Recognition Based On Image Processing And Multi-sensor Data Fusion

Atpresent, many industries have experienced no one charge of RMB banknotesidentification technology based on intelligent systems. Its core modules-notes identify the twomain functions of the system is to complete the different denominations of RMB clearingwork and identification of banknotes. Seeking for a simpler, rapid,having high recognitionrates banknote denomination recognition method and reliable to identify the authenticity ofbanknotes is the key to wide spreading the intelligent unmanned toll collection system.Firstly, For the Chinese market circulation of the fifth edition of RMB, The article isproposed based on image processing, real-time RMB identification method. This methodwill identify the characteristics of the RMB as an area. A comparative study with thetraditional RMB denomination identification method, a study found that in the Banknotespositioning, Freeman chain code method is improved and applied to the notes edge extractionhas good results. Then, improved the Freeman chain code identification method and applied iton the image edge extraction.Secondly, In order to further increase the reliability and adaptive ability of the currencydetector, multi-sensor are used to get more comprehensive data, a paper currencydiscrimination system of data fusion has been designed which based on fluorescence andmagnetic sensor. In the system, magnetic sensor, silicon photocell, filter and other moduleswere as hardware foundation. In the after analysis of the accepted signal in both time domainand frequency domain,and reasonable features are extracted, first of all to constructingimproved various radial base classification network for par value of the real co in, fusionhomologous feature level data, at the same time passive anti-fake method is used to class eachface value of paper currency and eliminate the false paper currency; Secondly, Classifierprobability of each sensor neural network as the follow-up to the frame of discernment basicprobability distribution; Finally, the D-S evidence theory is used for the secondDecision-making level data fusion, and getting the final recognition results. Finally, the result of experiments show that the improved Freeman chain code identificationmethod used in RMB banknote recognition, face recognition rate of95.625%,and it is aneffective image boundary extraction method. At the same time, Experiments show that thedata fusion algorithm and judging principle is effective, it is suitable for the the identificationof the authenticity of banknotes, and has high degree of accuracy, generalization ability andlower costs, and it has a strong practical value and significance of generalization.