(a)Typicalscreen-capturesofthevideo
clip
(b)
curve
(c) curve
Fig.9.
Gathering rstlyandthen
running.
(a)Typicalscreen-capturesofthevideo
clip
(b)
curve(c) curve
Fig.10.
Gatheringandrunningoccuratthesametime.
behaviorinlowandmediumcrowddensitywithlowcom-putationload.Moreaccurateestimationofthecrowddensityandcrowddistributionare rstlyintroducedintothedetectionofabnormalactivities.
However,theshortcomingofthethreshold-basedapproachisobvious:theappropriateestimationofthresholdiscrucialandsensitivetotheeffectiveness.Thethresholdalwayshavetobedeterminedbyexperience.Inthefuturework,wewillcombinethethreshold-basedmethodsandmachine-learning-basedmethodstogethertodetectmorekindsofabnormalbehaviors,suchasbelongingdropping,loitering,crossingoverthefence.Maybesomeappropriatemodelingapproachesandhardwareswithhighcomputationalpowersuchasgraphicsprocessingunit(GPU)willbeincludedin
500
ourfuturework.
REFERENCES
[1]L.KratzandK.Nishino,“Anomalydetectioninextremelycrowded
scenesusingspatio-temporalmotionpatternmodels”,inIEEECon-ferenceonComputerVisionandPatternRecognition,2009,pp.1446-1454.
[2]J.KimandK.Grauman,“Observelocally,inferglobally:Aspace-time
MRFfordetectingabnormalactivitieswithincrementalupdates”,inIEEEConferenceonComputerVisionandPatternRecognition,2009,pp.2921-2929.
[3]Y.Benezeth,P.-M.Jodoin,V.SaligramaandC.Rosenberger,“Abnor-malEventsDetectionBasedonSpatio-TemporalCo-occurences”,inIEEEConferenceonComputerVisionandPatternRecognition,2009,pp.2458-2466.
[4]H.Zhong,J.B.ShiandM.Visontai,“DetectingUnusualActivity
inVideo”,inIEEEConferenceonComputerVisionandPatternRecognition,2004,pp.2:819-827.
[5]E.L.Andrade,R.B.FisherandS.Blunsden,“DetectionofEmergency
EventsinCrowdedScenes”,intheInstitutionofEngineeringandTechnologyConferenceonCrimeandSecurity,2006,pp.528-536.[6]J.Yin,D.H.HuandQ.Yang,“Spatio-TemporalEventDetection
UsingDynamicConditionalRandomFields”,intheInternationalJointConferenceonArti cialIntelligence,2009,pp.1321-1327.
[7]X.Y.Wu,Y.S.Ou,H.H.QianandY.S.Xu,“ADetectionSystemfor
HumanAbnormalBehavior”,inIEEE/RSJInternationalConferenceonIntelligentRobotsandSystems,2005,pp.1204-1209.
[8]S.WangandZ.J.Miao,“AnomalyDetectioninCrowdScene”,in2010
IEEE10thInternationalConferenceonSignalProcessing,2010,pp.1220-1223.
[9]S.D.Wu,B.E.MooreandM.Shah,“ChaoticInvariantsofLagrangian
ParticleTrajectoriesforAnomalyDetectioninCrowdedScenes”,in2010IEEEConferenceonComputerVisionandPatternRecognition,2010,pp.2054-2060.
[10]J.S.Xie,L.Guo,Y.B.ChenandL.Zhao,“TheDetectionofUnusual

