手寫體數(shù)字識(shí)別系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)手寫體數(shù)字識(shí)別系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)中文摘要英文摘要設(shè)計(jì)方案系統(tǒng)設(shè)計(jì)流程預(yù)處理特征提取BP神經(jīng)網(wǎng)絡(luò)識(shí)別系統(tǒng)界面總結(jié)致謝手寫體數(shù)字識(shí)別系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)手寫體數(shù)字識(shí)別是多年來的研究熱點(diǎn)，也是字符識(shí)別中的一個(gè)重要問題。手寫體數(shù)字識(shí)別在郵政、金融領(lǐng)域應(yīng)用廣泛。當(dāng)涉及到數(shù)字識(shí)別時(shí)，人們往往要求識(shí)別器有很高的識(shí)別可靠性，特別是有關(guān)金額的數(shù)字識(shí)別時(shí)，因此，針對(duì)這類問題的處理系統(tǒng)設(shè)計(jì)的關(guān)鍵環(huán)節(jié)之一，就是設(shè)計(jì)出高可靠性和高識(shí)別率的手寫體數(shù)字識(shí)別方法。本文在對(duì)手寫識(shí)別主要方法分析和研究的基礎(chǔ)上，設(shè)計(jì)了一個(gè)基于BP神經(jīng)網(wǎng)絡(luò)的手寫體數(shù)字識(shí)別系統(tǒng)。該系統(tǒng)首先對(duì)數(shù)字字符進(jìn)行灰度化、二值化、細(xì)化、分割等預(yù)處理，然后采用逐象素特征提取法的特征提取方法提取字符特征，最后使用訓(xùn)練好的BP神經(jīng)網(wǎng)絡(luò)識(shí)別手寫的數(shù)字。實(shí)驗(yàn)表明，本系統(tǒng)對(duì)于較規(guī)范的手寫體數(shù)字的識(shí)別達(dá)到了很好的識(shí)別效果。DESIGNOFHANDWRITTENDIGHTRECOGNITIONSYSTEMHandwrittennumeralrecognitionisahotspotofstudyforyears,andisaimportantissueofcharacterrecognition.Handwrittennumeralrecognitionisappliedbroadlyinpostandfinancialenvironment.Whencomedowntonumeralrecognition,theemphasespeoplethinkisitsdependability,especiallyrefertomoney-digitrecognition.Sooneofthekeystepsforthesequestionsisdesigningahigh-dependabilityandhigh-accuracyhandwrittennumeralrecognitionsystem.Basedontheanalysisandresearchofthemethodofhandwrittenrecognition,thedesignofahandwrittennumeralrecognitionsystembasedonBPneuralnetworkisachievedinthispaper.Infirst,itisthepretreatmentofgraying,partition,thinningprogress,cutsapartetc.Thenalgorithmofpixel-by-featureextractionhasbeenadoptedinhandwrittennumeralrecognition.Finally,recognizenumberusingtheBPneuralnetworkwhichwaswelltrained.Experimentshowrecognitionsysteminthi