简介：Weproposeimprovedmultilevelfilters(IMLFs)involvingtheabsolutevalueoperationintothealgorithmicframeworkoftraditionalmultilevelfilters(MLFs)toimprovetherobustnessofinfraredsmalltargetenhancementtechniquesunderacomplexinfraredclutteredbackground.Comparedwiththewidelyusedsmalltargetenhancementmethodswhichonlydealwithbrighttargets,theproposedtechniquecanen-hancetheinfraredsmalltarget,whetheritisbrightordark.Experimentalresultsverifythattheproposedtechniqueisefficientandpractical.

简介：Calibratingasmallfieldcameraisachallengingtaskbecausethetraditionaltargetwithvisiblefeaturepointsthatfitthelimitedspaceisdifficultandcostlytomanufacture.Wedemonstrateanovelcombinedtargetusedincameracalibration.Thetangentpointssuppliedbyonecirclelocatedatthecenterofasquareareusedasinvisiblefeatures,andtheperspectiveprojectioninvarianceisproved.Bothvisibleandinvisiblefeaturesextractedbytheproposedfeatureextractionalgorithmareusedtosolvethecalibration.Thetargetsuppliesasufficientnumberoffeaturepointstosatisfytherequirementsofcalibrationwithinalimitedspace.Experimentsshowthattheapproachcanachievehighrobustnessandconsiderableaccuracy.Thisapproachhaspotentialforcomputervisionapplicationsparticularlyinsmallfieldsofview.

简介：Anewtechniqueispresentedtodetecttargetsunderseabackground.Focusedoninterestedareas,thespatialandfrequencydistributionsarefirststudied,thenanalyticexpressionoftargetsissetupbyleastsquaremethod(LSM).Accordingtotheshape,frequency,andanalyticexpressionoftargets,amatchedbiorthogonalwaveletisconstructedtobetterenhancetargetsandeliminatenoises.Resultsshowtheusefulnessofthismethodforsingleframedetection(signaltonoiseratioSNR≥1.25),whichprovidesabetterperformancethanclassicwaveletsandmorphologicalfiltering.