The Maximum Consensus Problem: Recent Algorithmic Advances

Outlier-contaminateddataisafactoflifeincomputervision.Forcomputervisionapplicationstoperformreliablyandaccuratelyinpracticalsettings,theprocessingoftheinputdatamustbeconductedinarobustmanner.Inthiscontext,themaximumconsensusrobustcriterionplaysacriticalrolebyallowingthequantityofinteresttobeestimatedfromnoisyandoutlier-pronevisualmeasurements.TheMaximumConsensusProblemreferstotheproblemofoptimizingthequantityofinterestaccordingtothemaximumconsensuscriterion.Thisbookprovidesanoverviewofthealgorithmsforperformingthisoptimization.Theemphasisisonthebasicoperationor""innerworkings""ofthealgorithms,andontheirmathematicalcharacteristicsintermsofoptimalityandefficiency.Theapplicabilityofthetechniquestocommoncomputervisiontasksisalsohighlighted.Bycollectingexistingtechniquesinasinglearticle,thisbookaimstotriggerfurtherdevelopmentsinthistheoreticallyinterestingandpracticallyimportantarea.