数字信号处理课件 第5章 Finite Impulse Response Filters.ppt

ChapterFive.FiniteImpulseResponseFilters1

ChapterFive.FiniteImpulseResponseFiltersananalogfilterwithanoisytoneinputandareduced-noisetoneoutput;thedigitalequivalentoftheanalogfilter.Ingeneral,filteringistheprocessingofatime-domainsignalresultinginsomechangeinthatsignalsoriginalspectralcontent.Thechangeisusuallythereduction,orfilteringout,ofsomeunwantedinputspectralcomponents.2

5.1.ANINTRODUCTIONTOFINITEIMPULSERESPONSE(FIR)FILTERSFIRdigitalfiltersuseonlycurrentandpastinputsamples,andnoneofthefilterspreviousoutputsamples,toobtainacurrentoutputsamplevalue.Givenafinitedurationofnonzeroinputvalues,theeffectisthatanFIRfilterwillalwayshaveafinitedurationofnonzerooutputvalues,Averagingthenumberofcars/minute.Thedashedlineshowstheindividualcars/minute,andthesolidlineisthenumberofcars/minuteaveragedoverthelastfiveminutes.3

WecanformalizethedigitalfilternatureofouraveragerbycreatingtheblockdiagraminFigure5-3showinghowtheaveragercalculatesitsoutputsamples.Inthegeneralcase,ifthekthinputsampleisx(k),thenthenthoutputisFigure5-3.Averagingfilterblockdiagramwhenthefifthinputsamplevalue,37,isapplied.4

Figure5-4.Alternateaveragingfilterstructure:inputvaluesusedforthe5thoutputvalue;inputvaluesusedforthe6thoutputvalue;inputvaluesusedforthe7thoutputvalue.5

5.2.CONVOLUTIONINFIRFILTERSFigure5-5.Averagingfilterconvolution:firstfiveinputsamplesalignedwiththestationaryfiltercoefficients,indexn=4;(b)inputsamplesshifttotherightandindexn=5;(c)indexn=6;fivecoefficientvalueswillbeindexedfromzerotofour,h(0)throughh(4).6

Figure5-5.Averagingfilterconvolution:(d)indexn=7;(e)indexn=8.Noticethatthetimeorderoftheinputshasbeenreversedfromtheinputsequenceorder.TheFIRfiltersimpulseresponseisi