基于機(jī)器視覺的孵化卵在線監(jiān)測(cè)系統(tǒng).pdf

1、合肥工業(yè)大學(xué)碩士學(xué)位論文基于機(jī)器視覺的孵化卵在線監(jiān)測(cè)系統(tǒng)姓名：何慧申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)系統(tǒng)結(jié)構(gòu)指導(dǎo)教師：史久根201204HatchingEggsOnlineMonitoringSystemBasedonMachineVisionABSTRACTThehatchingrateisaffectednotonlybythehatchingeggqualitywhethergoodorbad，butalsobythequalityo

2、fnewbornchicks，thefuturehealthofthepoultryandthepoweroflifeandperformanceTheactivitydetectionofembryoisanimportanttechnicalaspectintheegghatchingprocessInviewofthelaborintensive，lowefficiencyandpooraccuracyinmanualdetect

3、ionAnautomaticandpracticaldetectionsystembasedonmachinevisionsystemisdevelopedinsteadofmanualinspectionofhatchingeggforimprovingdetectingaccuracyandefficiencyThisdissertationfirstlyintroducestheresearchbackgroundanddevel

4、opmentofegghatchingactivityandmachinevisionThen，theegghatchingactivitydetectionhardwaresystemisbuilt，andthisdissertationanalysesthebasicconcepts，frameworkandkeytechniqueoftheimageenhancement，imagesegmentationandregionlab

5、eling，andthenthecolormodelisselectedFinally，thisdissertationpresentsthehatchingdetectionmethodThesimulatedannealingparticleswarmalgorithmandBPneuralnetworkaredescribedinadetail，andthenthekeytechniqueoftheneuralnetworkisf

6、urtheranalyzedandtheoptimalalgorithmisproposedtoimprovetheparametersofBPneuralnetworkThisdissertationstudiedfromseveralaspects，themainworkisasfollows：1Themachinevisionhardwaresystemisbuildforidentifyingexteriorqualityand

7、fertilityofhatchingeggThelightsourceandbackgroundcolorarefoundoutthroughalotofexperiments2TheeggsscreeningmethodsbasedonmachinevisiontechniquehasbeensystematicallystudiedColormodelandimagepreprocessingmethodslaythefoIund

8、ationforfurtherneuralnetworktraining3AnimprovedGASAisputforward，whichisusedtooptimizetopologystructureofBPneuralnetworkfordetectingfertilityofhatchingeggautomaticallyDetectionmethodoffertilityofhatchingeggduringtheentire

9、hatchingperiodisdevelopedsystematicallyHue、brightnessandsaturationfrequencyvaluesofthehatchingeggimageisselectedastheinputofneuralnetworkFertilityofhatchingeggisidentifiedbytheimprovedGA—SAneuralnetworkTheneuralnetworksy