AClusteringMethodforSpatio-temporalDataandItsApplication613strategiescanbefoundeveninonepasssequence;onecannaturallyrecognizeitfromthefactthatavideocameradoeszoom-upandzoom-outofagamescene.Inordertosolvetheseproblems,weemployedmultiscalematching[1],[2],apatternrecognitionbasedcontourcomparisonmethod.Andweemployedroughclustering[3],whicharesuitableofhandingrelativedissimilarityproducedbymultiscalematching.
Therestofthispaperisorganizedasfollows.Section2describesthedatastructureandpreprocessing.Section3describesmultiscalematching.Section4describesroughclustering.Section5showsexperimentalresultsontheFIFAworldcup2002dataandSection6concludestheresults.
2
2.1DataStructureandPreprocessingDataStructure
Weusedthehigh-quality,value-addedcommercialgamerecordsofsoccergamesprovidedbyDataStadiumInc.,Japan.Thecurrentstatesofpatternrecognitiontechniquemayenableustoautomaticallyrecognizethepositionsofballandplayers[4],[5],[6],however,wedidnotuseautomaticsceneanalysistechniquesbecauseitisstillhardtocorrectlyrecognizeeachactionoftheplayers.
Thedataconsistedoftherecordsofall64gamesoftheFIFAworldcup2002,includingbothheatsand nals,heldduringMay-June,2002.Foreachactioninagame,thefollowinginformationwasrecorded:time,location,names(number)oftheplayer,thetypeofevent(pass,trap,shootetc.),etc.Alltheinformationwasgeneratedfromthereal-timemanualinterpretationofvideoimagesbyawell-trainedsoccerplayer,andmanuallystoredinthedatabase.Table1showsanexampleofthedata.InTable1,’Ser’denotestheseriesnumber,whereaseries
Table1.Anexampleofthesoccerdatarecord
Ser
1
1
1
1
1
...
169TimeAction20:28:12KICKOFF20:28:12PASS20:28:12TRAP20:28:12PASS20:28:14TRAP22:18:42PENDT1SenegalSenegalSenegalSenegalSenegalP1T2P2X1Y1X2Y2100-3310Senegal190-50-1755019-1755019Senegal14-1221173004451143004451...France151440-685
denotesasetofcontiguouseventsmarkedmanuallybyexpert.Theremaining eldsrespectivelyrepresentthetimeofeventoccurrence(’Time’),thetypeofevent(’Action’),theteamID(’T1’)andplayerID(’P1’)ofoneactingplayer1,theteamID(’T2’)andplayerID(’P2’)ofanotheractingplayer2,spatial
Based on spatio-temporal correlation analysis of geographic time-series data, the paper proposes a kind of spatial clustering method which considers spatio-...
Databases, Data Mining, Spatio-temporal Databases...and its Application to Content and Request Routing...“Extraction and Clustering of Motion Trajectories ...
and Its Application in Virtual Mine System Wu ...It provides a way of thinking method and ...spatio-temporal data structure on the foundation ...
SPATIO-TEMPORAL DATA A... 暂无评价 6页 免费 Spatio-Temporal Data A... 暂无评价 4页 免费 The VGIST Multi-Modal ... 暂无评价 6页 免费 Bayesian ...
exploiting_temporal_gis_and_spatio_temporal_data_to_enhance_telecom_天文/地理_自然科学_专业资料。今日推荐 180份文档 2014证券从业资格考试...
Spatio-temporal Ontology for defining the quality of an application Béreng...its product specification, and data users to establish in what extent a ...
Spatio-Temporal Phrases for Activity Recognition_... which are created by clustering the local ... which is also a real application used in the ...
Multivariate Spatio-Temporal Clustering of Times-Series Data An Approach for...Application to Atmospheric Radiation Measurement Data Cluster analysis is a ...
we propose a spatio-temporal dependencies learning (STDL) method for action...STDL is an effective clustering technique for complex high-dimensional data. ...
Windowed_nearest_neighbour_method_for_mining_spatio-temporal_clusters_in_the_presence_of_noise.1_天文/地理_自然科学_专业资料。...
热门文档