Computer Vision：Models, Learning, and Inference

Thismoderntreatmentofcomputervisionfocusesonlearningandinferenceinprobabilisticmodelsasaunifyingtheme.Itshowshowtousetrainingdatatolearntherelationshipsbetweentheobservedimagedataandtheaspectsoftheworldthatwewishtoestimate,suchasthe3Dstructureortheobjectclass,andhowtoexploittheserelationshipstomakenewinferencesabouttheworldfromnewimagedata.Withminimalprerequisites,thebookstartsfromthebasicsofprobabilityandmodelfittingandworksuptorealexamplesthatthereadercanimplementandmodifytobuildusefulvisionsystems.Primarilymeantforadvancedundergraduateandgraduatestudents,thedetailedmethodologicalpresentationwillalsobeusefulforpractitionersofcomputervision.?Coverscutting-edgetechniques,includinggraphcuts,machinelearning,andmultipleviewgeometry.?Aunifiedapproachshowsthecommonbasisforsolutionsofimportantcomputervisionproblems,suchascameracalibration,facerecognition,andobjecttracking.?Morethan70algorithmsaredescribedinsufficientdetailtoimplement.?Morethan350full-colorillustrationsamplifythetext.?Thetreatmentisself-contained,includingallofthebackgroundmathematics.?Additionalresourcesatwww.computervisionmodels.com.