简介:Inthenormaloperationcondition,aconventionalsquare-rootcubatureKalmanfilter(SRCKF)givessufficientlygoodestimationresults.However,ifthemeasurementsarenotreliable,theSRCKFmaygiveinaccurateresultsanddivergesbytime.ThisstudyintroducesanadaptiveSRCKFalgorithmwiththefiltergaincorrectionforthecaseofmeasurementmalfunctions.Byproposingaswitchingcriterion,anoptimalfilterisselectedfromtheadaptiveandconventionalSRCKFaccordingtothemeasurementquality.Asubsystemsoftfaultdetectionalgorithmisbuiltwiththefilterresidual.Utilizingaclearsubsystemfaultcoefficient,thefaultysubsystemisisolatedasaresultofthesystemreconstruction.Inordertoimprovetheperformanceofthemulti-sensorsystem,ahybridfusionalgorithmispresentedbasedontheadaptiveSRCKF.Thestateanderrorcovariancematrixarealsopredictedbythepriorifusionestimates,andareupdatedbythepredictedandestimatedinformationofsubsystems.Theproposedalgorithmswereappliedtothevesseldynamicpositioningsystemsimulation.TheywerecomparedwithnormalSRCKFandlocalestimationweightedfusionalgorithm.ThesimulationresultsshowthatthepresentedadaptiveSRCKFimprovestherobustnessofsubsystemfiltering,andthehybridfusionalgorithmhasthebetterperformance.Thesimulationverifiestheeffectivenessoftheproposedalgorithms.
简介:分析了几种数字梳状滤波器在进行视频信号的亮、色分离时的优、缺点。针对现有的1D、2D数字梳状滤波器的局限性,通过引入图像前后场内容的相关性,实现了基于图像内容的BD运动自适应数字梳状滤波器的算法。通过与实际图像比对,证明了采用这种处理方法的滤波器能够达到非常好的亮、色分离效果。