简介:Inthispaper,wepresentahighspeedautofocussystemformicrosystemapplicationsanddesignalook-up-tablebasedautofocusingalgorithmforapplicationswhenatargetobjectisalwaysvisible,e.g.,manufacturingpartswithalignmentfiducials.Weperformanevaluationof24focusmeasurestoverifythatwhichfocusmeasureisthebestforthelook-up-tablebasedmethod.Fromtheevaluation,wefindthattheChebyshevmoments-basedfocusmeasure(CHEB)isthemostsuitable.Furthermore,wealsodevelopalook-up-tablebasedautofocussystemthatusesCHEBasthefocusmeasure.Intrainingphase,weofflineconstructatablefromtrainingimagesofanobjectthatarecapturedatseverallensdistances.Eachentryoftableconsistsoffocusmeasurecomputedfromimageandlensdistance.Inworkingphase,givenaninputimage,thealgorithmfirstcomputesthefocusmeasureandthenfindsthebestmatchfocusmeasurefromthetableandlooksupthecorrespondinglenspositionformovingitintothein-focusposition.Ouralgorithmcanperformautofocusingwithinonly2stepsoflensmoving.Theexperimentshowsthatthesystemcanperformhighspeedautofocusingofmicroobjects.