基于Adaboost算法的人脸检测研究及实现(3)
发布时间:2021-06-06
发布时间:2021-06-06
基于Adaboost算法的人脸检测研究及实现
Abstract
Title:STUDYANDIMPLEMENTAl-lONOFFACEDETECTl0NBASED
ONADABoOSTALGORITHM
Major:SignalandInformationProcessing
Name:PengLIN
Supervisor:Prof.ErHuZHANGSignature:让s’9舱ture.丝彳
asaAbstractFacedetectioninpattemrecognitionandmachinevisiontechnology
valuelong-termrapidsubjectofconcern,hashi曲academicandcommercialvalue.Withthe
developmentofintelligentcomputingtechnology,newmethodsandnewtechnologieswill
continuetobeused,whichbringmorevitalitytothe
increasinglybetterresearch,anddetectionresultwillbeandfaster.
Theresearchstartedfromtheadaboostalgorithm,whichwasoptimizedfromthetraining
processandthedetectionprocess,andOilthisbasisfacedetectionandtrackingsystemwas
developed.
Firstly,basingontheclassicalprinciplesoftheadaboostalgorithm,thetrainingprocess
andthedetectionprocesswereachieved
arespectivelyandindependently.Secondly,inandtheviewofthetrainingprocessconsumedsuch
trainingprocesswerelongtime,thenumberoffeaturesthattrainingtimewasstepsoftheoptimizedrespectively,soreducedbynearly50%.
Then,fromthepointofspeedrequiredbyreal—timedetectionsystem,afacetracking
algorithmbasedontheadaboostalgorithmwasproposed,andtheaverageprocessingtimeof
singlefianlewasshortenedsharply.Finally,afterrealizingthedetectionandtracking
algorithm,thefacedetectionandtrackingsystemwasdeveloped,includingimage—basedface
detectionsubsystem,video—basedfacedetectionandtrackingsubsystemandvideo
surveillancesubsystem,Inaddition,byembeddedofthecomplexionmoduleandtheprincipal
componentanalysismodule,theperformanceofthesystemwasgreatlyimproved.
andimprovingtheBecauseofrealizingtheadaboostalgorithmindependently,therelevantprocedurescouldbetran印lantedeasily,andagoodfoundationwaslayforf=Ilrtherstudy
application.
KeyWords:Facedetection;Adaboostalgorithm;Facetracking;Systellaoptimization
下一篇:中专学校新生报名登记表