Study on Recognition of Coal and Gangue Base on Di

LiuGangfengStudyonRecognitionofCoalandGangueBaseonDigitalImageProcessingTechnologyReportContentBackgroundResearchPurposeRecognitionsystemdesignResearchonFastAlgorithmofRecognitionBackgroundGangue:Producedintheprocessofcoalminingandwashing，andtobestoredasganguemountain.Largenumberofgangueinthecoalwillreducethecoal’sgrade,ganguemountainsoccupylotsoffarmlandandleadtoenvironmentpollution.ResearchAimBasingonthegraydifferencebetweencoalandgangue,weplantodesignanauto-detectionsystem.Toprovideatechnicalproposalforseparatinggangueundertheshaft.auto-separationsystemofgangueQueuingSystemofgangueflowrecognitionsystemofgangueseparationsystemelectronic-controlsystemrecognitionsystemofganguerecognitionsystemofgangueisthecorepartofauto-separationsystemofgangue,thissystem’smainfunctionistorecognizetheganguefromcoal,andsendtheresulttoseparationsystemrecognitionsystemismadeupofthreeparts:illuminationsystem,CCDandbackboard.●imagegrayscale:theprocessoftransitiongrayimagefromcolorimage.●Y=0.299R+0.587G+0.114B。●Y:luminance，R:red,G:green,B:Blue。imagegrayscalegrayHistogramofimageGrayHistogramistodescribethenumberofpixelinallthegrayvalue，Butitcan’tdescribethecoordinateofeverypixelintheimage,becausethespacemessagehavebeendropped.imagebinaryzationImagebinaryzationistotransformimageintobinaryimageusingonedesignatedthreshold,whichmeansthatallthepixelpointiseither0or255,aftertreatment,theimagebecomesablack-and-whitepicture.Afterbinaryzationtreatment,theimageonlyhasthepointwhosegrayvalueis0or255,butthistypeofimagealsocanreflectthewholeimageshapeandthepartedgefeature.Theeffectofbinaryzationisrelatedtothesizeofthreshold.TyxfTyxfyx,255,0,gTyxfTyxfyx,255,0,gHowtochoosethresholdDifferentthreshold：T=70；T=100；T=130。CCD摄像头CCDcamera:highrate,highresolution,highimagequality.ThissubjectchoosesMV-VD036SMCCDMV-VD036SMCCD:Signal-to–NoiseRatiois64dB,timeofexposureis1/10000sto1s,thehighestframerateis60picturespersecond.EffectofbackboardonrecognitionsystemTherecognitionofgangueisbasedonthehypothesis:backgroundisuniformandhaslargedifferencewithgangueingrayvalue,thenwecanextractthegangueandcoal’simagefromthewholepictureusingthresholddivision.Thegrayleveldistributionofordinarybeltisbetween100and200.Thebelt’sgrayareinstable,causedbythedifferentcomponentandbelt’smoving.HowtofixthebackboardHowtochoosebackboardTheperfectbackboardisdifferentfromcoalandgangueinthegray’sdistribution.Thecoalandgangue’sgraydistributesinbetween50and200underproperlightintensity,sotheprefectbackboard’sgrayisdistributedinhighgrayareaorlowgrayarea.Therearetwotypebackboard:oneisater-board,whichhavealessreflectivity,theotheriscoloredboardorwhiteboard,whichissensitivetothelight.ater-boardandcoal’sgraydistributionater-boardandgangue’sganguedistributionconclusionater-board’sgrayismainlydistributedaround50Underthisbackboardtocollectcoal’simage,it’sdifficulttoseparatethecoalfrombackboard,becauselotsofcoal’sgraydistributionisthesamewithbackboard.Gangue’sgraydistributionisaround100,sotherearelargedifferencesbetweengangueandbackboard,wecanseparategangueinthreshold.graydistributionofcoal,gangueandwhitebackboard,problemThewhitebackboard’sgrayisdispersiveandsensitivewithlightstrength.Thewhitebackboardiseasypollutedbydust,itsgraywillreduceandinfluencethereorganizationresult.Intheprocessofcollectimage,itisdifficulttoavoidtheshadow’sproductionbycoalorgangueonthebackboard.Theshadowwillinfluencetheimagegray.influenceoftimeonlightstrengththeinfluenceoflightstrengthongrey15LED30LED45LEDGangue,coal,andbackground‘sgraycomparisonganguerecognitionflowCalculatethreshold’smaximumvariance•Wedivideallpixelsintotwocategories,andthenfindthemaximumvariancebetweenthetypes,todeterminethethreshold.•ifgraylevelkisthethresholdtodivideimageintotwoparts:thepixelswhosegraylevelis0~kareAcollectionandtheother(k+1~255)belongingtocollectionB.)(1)()]()([)()(2222kkkkBBAAsystemparameterofimagecollectionCCDsamplingfrequency：30HzGangue’sspeed：2m/slightintensity：45LEDBackboardgray:25-98，averagegray54。（1）collectimage（2）differenceimageImagedividedwithmaximumvariancethresholddifferenceimage（3）binaryzation（4）calculatebinaryimage’sarea•binaryimage’sareaisthenumberofpixelwhosepixelis255.•Coal/gangue’sareaare:20910，28011，28072，10828。•theimage’sformis452×400,it’smeanthateveryimagehas180800pixels。•Ifwesettheareathresholdis21000，imageaanddwillbeconsideredtherearenocoalorgangue.（5）calculatetheaveragegrayvaluefirst,thissubjectdesignateathresholdtobethecut-point，calculatethesumofallthepixelpointwhosegrayvalueisgreaterthanthecut-point,thendividethenumberoffore-imagepixeltogettheaveragegrayvalue.IdesignaprogramwithMatlab,thentocalculatethesefourimage，theiraveragegrayvalueare：74.5548，73.6866，72.2348，67.1110。RecognitionofpollutedgangueAllwedobeforeisbasedonthegangueisclean,it’smeanthattherearenocoalparticleonthesurfaceofgangue,butit’sdifficulttorealizeinpractice.Sothissubjectneedtohaveanexperimentforrecognitiontheganguepollutedbycoal.PollutedgangueThenumb