Font Size: a A A

Research On Image Processing System Of Bill Sorter

Posted on:2013-07-29Degree:MasterType:Thesis
Country:ChinaCandidate:C LiFull Text:PDF
GTID:2268330401450773Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
The image processing technology and pattern recognition of bills is the keytechnology of the bill sorter,and they are important means to improve sorting speedand efficiency. In view of the situation that recognition algorithm is still not matureenough for application,and there is not satisfactory to achieve the result in the sorter.In order to improve effectively of the sorter, in this paper, the rapidly recognition rate,portability, and real-time performance characteristics of the bills recognitionalgorithm and the corresponding hardware circuit is designed.In this paper, a system circuit including a DSP processor and a CPLD controlleris designed.It has selected CIS sensor to collect bills images and used A/D conversionand data buffing,and then has stored datas in the extended off-chip memory of thesystem. After that the systerm has used DSP to processe the datas.CPLD devices as alogic control and timing-driven chip, to control the CIS sensors can worknormally,and in conjunction with the A/D converter to generate the appropriate timingof working.A series of image preprocessing of the bills are carried out.It has introduced thecommon two kinds of noise and has introduced the method to remove the noise in theimage. Considered the deviation comprehensivily when detecting the edge and fittinga straight line, It has used edge detection method for an image edge damaged andangle crippled.and improved a least squares method of fitting a straight line accuratelyand quickly.There are about five recognition algorithm of bills in the clear extension. Amongthem, when identified the serial number of bills,It has proposed a charactersegmentation algorithm that used bidirectional shortest path to search character profilecurve bump position as the cutting point,and using a the method to identify thecharacter. Through the simulation experiment, the correct recognition rate can reachmore than90%. Introduced the template matching method to identify thedenominations of bills.Because this method have drawback that is difficult to meet therequirements of the system real-time,proposed a simple method that repair rectanglearea to identify the denominations of bills which can efficiently identify all the imagesof the bills.Using the different regions of image grayscale have significant differencecharacteristics in bills to distinguishing the positive and negative direction of the bills.According to incomplete in the area of the whole note in proportion,to judge the notes are available. In order to improve the degree of the bill recognition of the correct rate,the article have used a method that is to contrast the entire average gray value of billsto the corresponding template gray value bills for the ratio,then judged the old andnew grades based on the ratio.In order to verify the effectiveness and real-time of the algorithms in the systerm,the arcticle have designed a interface of bill recognition system which based onMatlab/GUI.Then simulated the algorithms in the recognition system which involvedin several identifiable part of the bill. From the simulation results, bills recognitionalgorithm can meet clear subsystems requirements in the paper money processingsystems.
Keywords/Search Tags:image processing of bills, pattern recognition, bill sorter, Matlab/GUI
PDF Full Text Request
Related items