結合小波域去噪和運動估計的監(jiān)控場景分析研究.pdf

1、河北工業(yè)大學碩士學位論文結合小波域去噪和運動估計的監(jiān)控場景分析研究姓名：田曉紅申請學位級別：碩士專業(yè)：通信與信息系統(tǒng)指導教師：于明20091101結合小波域的去噪和運動估計的監(jiān)控場景分析研究iiSURVEILLANCESCENEANALYSISUSINGDENOISINGMOTIONESTIMATIONINWAVELETDOMAINABSTRACTWiththedevelopmentofvideomonitingtechnologypr

2、ocessingofsurveillancevideohasbecomearesearchhotspot.Surveillancevideooftencontainssomenoiseduetochangesinweatherlowlightsoon.Noisenotonlywsenvideoqualityaffectpeople’ssubjectivefeelingbutalsoenlargethetransmissionbwidth

3、datastagespacethatbringingalotofinconveniencesffurtherprocessing.Inadditionmonitingsceneistempallystationaryvideosequenceshasstrongtempalcrelationthedatageneratedbycameraareverylargesoincreasesdatatransferstagecosts.Tore

4、solvetheseproblemsthispapermainlyresearcheswaveletdenoisingmotionestimationinmonitingscene.Traditionalvideodenoisingmethodscandebaseimagenoisebutdestroyimageedgeinfmation.Wavelettransfmhasgoodtimefrequencyacteristicshasb

5、ecomeanefficientmathematicstoolappliedtoimagedenoising.Accdingtothenoisetypeinsurveillancevideothepapertakesadvantagesoftheacteristicsofwaveletthresholddenoisingmethodwienerfilteringproposeanimagedenoisingalgithm.Compare

6、dwithwavelettransfmthresholddenoisingalgithmsspatialwienerfilteringtheproposedmethodcanimprovePSNRefficiently.Videosequenceinmonitingsceneisdifferentfromtheheadshoulderimagesequencemovingobjectsmayoccupyonlyasmallpartoft

7、hewholeimage.Ifwedomotionestimationftheentireimageitwillnotonlytimeconsumingbutalsolabious.Thispaperproposesanalgithmthatapplyingmovingobjectdetectionintowaveletdomaincombiningwithmotionestimationthatreducingunnecessarys

8、earchsavingstagespace.Theexperimentalresultsprovethatthisapproachdecreasethecomputationalcomplexityobviouslycomparewithwaveletmotionestimationblockmatchingalgithm，thePSNRcanbeimproved3.077dB.KEYWDS:videomonitingvideodeno