X射線智能評片系統(tǒng)的研究.pdf

1、湖南大學(xué)碩士學(xué)位論文X射線智能評片系統(tǒng)的研究姓名：賀繼德申請學(xué)位級別：碩士專業(yè)：檢測技術(shù)與自動(dòng)化裝置指導(dǎo)教師：劉波峰20080320x射線智能評片系統(tǒng)的研究AbstractJu吣ingX—rayfilm’squalityis觚iInponantpaniIlX—raydetectionTraditionally，thejudgingworkisdonebyjudgingworker，but“，saverylargeworkIthas111

2、anydisadVantages，forexample，stronglightwillhurttheeyes，lowemciency，andthedefectionofthetechnicalstaffa齬essedthequality，eXl)erienceandtheimpactofextemalcoIlditions，resultsoRenVary矗ompersontopersonAnificail—decfectioncanno

3、tmeetgrowingdevelopmemneedSW曲therapiddeVelopmentofcoInpupters，d逸itialirnageprocessingtechnologies，d逸“alfilm，i11telligentdetectionhasbecomeanondestrctivedirectionIntheX—rayfilmtoSohrethed遠(yuǎn)italissue，theiIllagedefectioniden

4、tificationandclassificationoffilmsonasmanasSessmentbecoIT圮apriorityThepaper∞ekstobuildai11tell遠(yuǎn)entidentificationsystem，iIltheimageoftheknowledgebaseconstmctionondefectsandde位ienciesidentifiedclassi&atiOnreasoniIlgdonea10

5、tofworkonttlestudy111roughInuchdatacollection，theknowledgebaseofknowledgeiIlthispaperisdividedilltotwoparts，shallowl【llowledgebaseaIlddeepl(IlowledgebaseShaIlowknowledgeba跎isastandardfuzzyBPneuralne附orl(，l(1lowledgethati

6、siIIlplicitwithiIl，療omtheshapeofiIIlagedefectsanddeficienciesiInagecatego吼c鋤beobtailledtIlroughtheleamiIlgnetworkrelatedparaImters蛐chastherightmatri】【andVectortIlresholdDeepknowledgebaseisre丑ectediIl觚oaSpects：First，the∞l

7、fknowledge，theleamiIlgisnotpassedd打ecny丘omtheeXperienceoftheacquisitionofl【IlowledgeDefectsiIlthepattemrecognitionfllzzyonthistechnologyandanificialneuralneMorktechn0109y，∞1ectfuzzyBPneuralne押orklearn證ga培orithmThroughthe

8、simulationtestandprovethat∞uralne錒orl【sandfuzzyBPl妒ne倆orkshoncomiIlgsoftraditionalrecognitionrate罄∞nlparedtothe凡zzytheorytraillmgsampless0thatthemappingne倆orkcapac“yandimprovetheabnityofthe船倆orkTllerefore，theabilitytoide

9、ntif)rtheformerthanthelatterismuchhigher、^rIlentheshallowknowledgebasecannotbegiventhecorrectidemi丘cation(idemificationofUnCerta砷y)，thesystemthrou曲theactivationofdeepl【IlowledgebaSeforreasoIli】唱，C0metecognizetheresultsIn

10、ferenc跫agaillSttheiIlforImtioniIlthedatabase，∞lectediIlthel【Ilowledgebaseonthecurrentseamofuse旬lknowle姑ereaSonillgjudgementInthispaper，des遠(yuǎn)Iledbyd遠(yuǎn)italXrayfilmdefectsil呲elligentidentificationsystemafterala唱enumberofsimul