基于场景自适应收敛非均匀性校正方法(英文)

基于场景自适应收敛非均匀性校正方法(英文)Adaptive Convergence Non-Uniformity Correction Method Based on Scene for Target

() 基于场景自适应收敛非均匀性校正方法英文 AdaptiveConvergenceNon-UniformityCorrectionMethod BasedonSceneforTargetDetection Abstract： Targetdetectioniswidelyusedinmilitary,industrial,and environmentalfields.However,thenon-uniformityoftheimaging sensorbringschallengestotargetdetection.Thetraditional non-uniformitycorrection(NUC)methodcanonlycalibratethe non-uniformityunderacertainscene,whichcannotguaranteethe correctionunderotherscenes.Inthispaper,anadaptive convergencenon-uniformitycorrectionmethodbasedonsceneis proposed.Firstly,theimagingsensorismodeledasatime-variant system.Then,afiniteimpulseresponse(FIR)filterisdesigned basedonminimizingthemeansquareerror(MSE).Finally,the proposedmethodisevaluatedontheinfraredimagedataset. Keywords：targetdetection,non-uniformitycorrection, adaptiveconvergence Introduction： Withtherapiddevelopmentofimagingtechnology,target detectionhasbecomeanimportantapplicationinmanyfields, suchasmilitary,industrial,andenvironmentalfields.However,the imagingsensor'snon-uniformitybringschallengestotarget detection.Theimagingsensor'snon-uniformitycancausespatial patternsthatareunrelatedtothescene,whichdegradesthe imagequalityandaffectsthetargetdetectionperformance.The non-uniformitycorrection(NUC)methodisusedtocalibratethe non-uniformityoftheimagingsensor.ThetraditionalNUC methodsrelyonthecalibrationdata,whichisobtainedbeforethe imagingsensorisused.However,thenon-uniformitychanges underdifferentscenes,whichmeansthecalibrationparameters needtobeadjustedaccordingtothescene'scharacteristics.The traditionalNUCmethodcannotguaranteethecorrectionunder otherscenesthataredifferentfromthecalibrationscene. Therefore,anadaptiveconvergencenon-uniformitycorrection methodbasedonsceneisproposedinthispaper. Methodology: