中国数字书法检索与作品真伪鉴别的研究

《中国数字书法检索与作品真伪鉴别的研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、浙江大学博士学位论文中国数字书法检索与作品真伪鉴别的研究姓名：章夏芬申请学位级别：博士专业：计算机应用指导教师：庄越挺20061001浙江大学博士学位论文5．系统实现。本文实现了上述数字书法检索和作品真伪鉴别的理论、技术和方法，并给出其在中美百万册数字图书馆中的实际运行结果。关键词数字图书馆中国书法页面切分特征提取形状匹配书法字检索书法风格真伪鉴别lI浙江大学博士学位论文AbstractWithrapiddevelopmentofdigitizingtechnology,thedigitizationofChinesehistoricalcalligraphyworksism

2、archingintoanewstage。Theoriginalhistoricalcalligraphyworksareexistsinpapers，stones,silksorbambooslips．Theyareunique。fragile，andc翘beeasilydestroyed．Thedigitizinggivesthemasecondarymediumtokeepandtoenableuniversalaccessible．Butdigitizingthehistoricalcalligraphyworksbringsnewchallenges：Astheex

3、istingOpticalCharacterRecognitiontechnologiescarl’tconvertthesecalligraphyimagesintotexts．howtoretrievethecalligraphycharactersandenablethesearchingservicesindigitallibrary?Ascalligraphyworks’verificationhaslongbeeninvestigatedbycalligraphistssubjectivelyfromtheart’sfield．Howtodistinguishth

4、egenuinecalligraphyworksfromthoseofforgedwithobjectivemeasupesandevidences?Thisthesisdiscussestheories，methodologiesandtechnologiesofhowtomtrievecalligraphyworksandhowlobringobjectivemeasurestoverifythehistoricalcalligraphyworks，Themaincontributionsofthisthesisareasfollows：1．Retrievalofshap

5、e-basedcalligraphycharacter．Chapter3introduceshowtoextractmeaningfulfeaturesfromindividualcalligraphycharacterimagessegmentedfromtheca

6、ligraphyimagepage．Chapter4proposesapproachesofhowtoretrievecalligraphycharacterswithoutrecognition．Fi嫩shapematrixisconstrucedtforeachcalligraphycharacterbas

7、edonthecoutourpoint’sproperty．Thenretrievethecandidatesimilartothequeryintermsofshape．Meanwhile，ifthedatabasecontainsnocalligraphycharacterthatmeetstheusar’sdemands，thenailewstyleofcalligraphycharacterwillbegeneratedbylearningfromtwosimilarstyles。2。Fastsearchengineofcalligraphycharacter．The

8、responsetimeisimportantfortheserviceofdigitallibrary；andsoChapter5proposes3sWategestospeeduptheretrievaltime。Thefirstistouseacoarse-to-fineprocesstoshortenthetotalcomputationtime．ThesecondistoemploygoodalgorithmsoftheDyanamicprojectinghistogramwarpingand