眼睛识别.docx

眼睛识别.docx

《眼睛识别.docx》由会员分享，可在线阅读，更多相关《眼睛识别.docx（16页珍藏版）》请在冰豆网上搜索。

眼睛识别

AbouttheAuthor

zafersavas

Engineer

ASELSANA.S.

 Turkey

Member

ZaferisanelectronicsengineerlivinginAnkara/TurkeyandworkingforASELSANA.S./TMM. 

Hehasbeencodingforabout12years,andcan'tthinkalifewithoutC/C++.

Helikes

-MSVisualC++6.0andMFC

-.NETC#

-Allkindsofelectronicsstuff

-MachineVisionprojects

Alsoheenjoys 

-Listeningto"TheCranberries"&"TheGloriousDolores"

-Travelling

E-mail:

zafersavas@

Introduction

Eyesarethemostimportantfeaturesofthehumanface.Soeffectiveusageofeyemovementsasacommunicationtechniqueinuser-to-computerinterfacescanfindplaceinvariousapplicationareas.

Eyetrackingandtheinformationprovidedbytheeyefeatureshavethepotentialtobecomeaninterestingwayofcommunicatingwithacomputerinahuman-computerinteraction（HCI）system.Sowiththismotivation,designingareal-timeeyefeaturetrackingsoftwareistheaimofthisproject.

Thepurposeoftheprojectistoimplementareal-timeeye-featuretrackerwiththefollowingcapabilities:

∙RealTimefacetrackingwithscaleandrotationinvariance

∙Trackingtheeyeareasindividually

∙Trackingeyefeatures

∙Eyegazedirectionfinding

∙Remotecontrollingusingeyemovements

InstructionstoRunandRebuildTrackEye

InstallationInstructions

1.Extract TrackEye_Executable.zip file.Beforerunning TrackEye_636.exe,copythetwofiles SampleHUE.jpg andSampleEye.jpg tothe C:

\ folder.ThesetwofilesareusedforCAMSHIFTandTemplate-Matchingalgorithms.

2.Therearenootherstepstobefollowedbytheusertorunthesoftware.TherearenoDLLdependenciesasthesoftwarewasbuiltwiththeDLLsstaticallyincluded.

SettingstobeDonetoPerformaGoodTracking

SettingsforFace&EyeDetection

UnderTrackEyeMenu-->TrackerSettings

∙InputSource:

video

∙ClickonSelectfileandselect ..\Avis\Sample.avi

∙FaceDetectionAlgorithm:

HaarFaceDetectionAlgorithm

∙Check“TrackalsoEyes”checkBox

∙EyeDetectionAlgorithm:

AdaptivePCA

∙Uncheck“VarianceCheck”

∙NumberofDatabaseImages:

8

∙NumberofEigenEyes:

5

∙Maximumallowabledistancefromeyespace:

1200

∙Facewidth/eyetemplatewidthratio:

0.3

∙ColorSpace typetouseduringPCA:

 CV_RGB2GRAY

SettingsforPupilDetection

Check“Trackeyesindetails”andthencheck“Detectalsoeyepupils”.Click“AdjustParameters”button:

∙Enter“120”asthe“ThresholdValue”

∙Click“SaveSettings”andthenclick“Close”

SettingsforSnake

Check“Indicateeyeboundaryusingactivesnakes”.Click“Settingsforsnake”button:

∙Select ColorSpace touse:

 CV_RGB2GRAY

∙SelectSimplethresholdingandenter100asthe“Thresholdvalue”

∙Click“SaveSettings”andthenclick“Close”

Background

Sofartherehasbeenalotofworkoneyedetectionandbeforetheproject,thepreviousmethodswerecarefullystudiedtodeterminetheimplementedmethod.Wecanclassifystudiesrelatedtoeyeintotwomaincategoriesaslistedbelow:

SpecialEquipmentBasedApproaches

Thesetypeofstudiesusethenecessaryequipmentwhichwillgiveasignalofsomesortwhichisproportionaltothepositionoftheeyeintheorbit.VariousmethodsthatarecurrentinuseareElectrooculography,Infra-RedOculography,Scleralsearchcoils.Thesemethodsarecompletelyoutofourproject.

ImageBasedApproaches

Imagebasedapproachesperformeyedetectionsontheimages.Mostoftheimagebasedmethodstrytodetecttheeyesusingthefeaturesoftheeyes.Methodsusedsofarareknowledge-basedmethods,feature-basedmethods（color,gradient）,simpletemplatematching,appearancemethods.Anotherinterestingmethodis“Deformabletemplatematching”whichisbasedonmatchingageometricaleyetemplateonaneyeimagebyminimizingtheenergyofthegeometricalmodel.

ImplementationofTrackEye

Theimplementedprojectisonthreecomponents:

1.Facedetection:

Performsscaleinvariantfacedetection

2.Eyedetection:

Botheyesaredetectedasaresultofthisstep

3.Eyefeatureextraction:

Featuresofeyesareextractedattheendofthisstep

FaceDetection

Twodifferentmethodswereimplementedintheproject.Theyare:

1.ContinuouslyAdaptiveMeans-ShiftAlgorithm

2.HaarFaceDetectionmethod

ContinuouslyAdaptiveMean-ShiftAlgorithm

AdaptiveMeanShiftalgorithmisusedfortrackinghumanfacesandisbasedonrobustnon-parametrictechniqueforclimbingdensitygradientstofindthemode（peak）ofprobabilitydistributionscalledthemeanshiftalgorithm.Asfacesaretrackedinvideosequences,meanshiftalgorithmismodifiedtodealwiththeproblemofdynamicallychangingcolorprobabilitydistributions.Theblockdiagramofthealgorithmisgivenbelow:

Haar-FaceDetectionMethod

ThesecondfacedetectionalgorithmisbasedonaclassifierworkingwithHaar-Likefeatures（namelyacascadeofboostedclassifiersworkingwithHaar-likefeatures）.Firstofallitistrainedwithafewhundredsofsampleviewsofaface.Afteraclassifieristrained,itcanbeappliedtoaregionofinterestinaninputimage.Theclassifieroutputsa"1"iftheregionislikelytoshowfaceand"0"otherwise.Tosearchfortheobjectinthewholeimage,onecanmovethesearchwindowacrosstheimageandcheckeverylocationusingtheclassifier.Theclassifierisdesignedsothatitcanbeeasily"resized"inordertobeabletofindtheobjectsofinterestatdifferentsizes,whichismoreefficientthanresizingtheimageitself.

EyeDetection

Twodifferentmethodswereimplementedintheproject:

1.Template-Matching

2.Adaptive EigenEye Method

Template-Matching

Template-Matchingisawell-knownmethodforobjectdetection.Inourtemplatematchingmethod,astandardeyepatterniscreatedmanuallyandgivenaninputimage,thecorrelationvalueswiththestandardpatternsarecomputedfortheeyes.Theexistenceofaneyeisdeterminedbasedonthecorrelationvalues.Thisapproachhastheadvantageofbeingsimpletoimplement.However,itmaysometimesbeinadequateforeyedetectionsinceitcannoteffectivelydealwithvariationinscale,poseandshape.

AdaptiveEigenEyeMethod

Adaptive EigenEye Methodisbasedonthewell-knownmethod EigenFaces.Howeverasthemethodisusedforeyedetectionwenameditas“EigenEye Method”.Themainideaistodecomposeeyeimagesintoasmallsetofcharacteristicsfeatureimagescalledeigeneyes,whichmaybethoughtofastheprincipalcomponentsoftheoriginalimages.Theseeigeneyesfunctionastheorthogonalbasisvectorsofasubspacecalledeyespace.Howeverweknowthatthe eigenface methodisnotscaleinvariant.Toprovidethescaleinvariancewecanresizetheeye-databaseoncewiththeinformationgatheredbythefacedetectionalgorithm（EyeWidth/FaceWidth?

0.35）,wecanprovidescale-invariantdetectionusingonlyonedatabase.

OpenCVFunctionsforObjectTrackingandDetection

OpenCV Libraryoffersalotofimageprocessingandobjecttracking&detectionlibraries.Themainfunctionusedintheseprojectsandtheirusagearegivenbelow:

SampleCodeforHaar-FaceTracking

 Collapse | CopyCode

voidCTrackEyeDlg:

:

HaarFaceDetect（IplImage*img,CvBox2D*faceBox）

{

intscale=2;

IplImage*temp=cvCreateImage（cvSize（img->width/2,img->height/2）,8,3）;

CvPointpt1,pt2;

inti;

cvPyrDown（img,temp,CV_GAUSSIAN_5x5）;

#ifdefWIN32

cvFlip（temp,temp,0）;

#endif

cvClearMemStorage（storage）;

if（hid_cascade）

{

CvSeq*faces=cvHaarDetectObjects（temp,hid_cascade,storage,1.2,2,

CV_HAAR_DO_CANNY_PRUNING）;

NumOfHaarFaces=faces->total;

if（NumOfHaarFaces>0）

{

CvRect*r=（CvRect*）cvGetSeqElem（faces,0,0）;

pt1.x=r->x*scale;

pt2.x=（r->x+r->width）*scale;

#ifdefWIN32

pt1.y=img->height-r->y*scale;

pt2.y=img->height-（r->y+r->height）*scale;

#else

pt1.y=r->y*scale;

pt2.y=（r->y+r->height）*scale;

#endif

faceBox->center.x=（float）（pt1.x+pt2.x）/2.0;

faceBox->center.y=（float）（pt1.y+pt2.y）/2;

faceBox->size.width=（float）（pt2.x-pt1.x）;

faceBox->size.height=（float）（pt1.y-pt2.y）;

}

}

cvShowImage（"Tracking",img）;

cvReleaseImage（&temp）;

}

SampleCodeforCamShiftAlgorithm

 Collapse | CopyCode

//InputsforCamShiftalgorithm

IplImage*HUE=cvCreateImage（cvGetSize（SampleForHUE）,IPL_DEPTH_8U,1）;

extractHUE（SampleForHUE,HUE）;//**ExtractHUEinformation

inthist_size=20;

floatranges[]={0,180};

float*pranges[]={ranges};

hist=cvCreateHist（1,&hist_size,CV_HIST_ARRAY,pranges,1）;

cvCalcHist（&HUE,hist）;//CalculatehistogramofHUEpart

hueFrame=cvCreateImage（cvGetSize（CameraFrame）,IPL_DEPTH_8U,1）;

backProject=cvCreateImage（cvGetSize（CameraFrame）,IPL_DEPTH_8U,1）;

extractHUE（CameraFrame,hueFrame）;

while（trackControl!

=0）

{

extractHUE（CameraFrame,hueFrame）;

cvCalcBackProject（&hueFrame,backProject,hist）;//Probabilityisformed

//cvShowImage（"Tester2",backProject）;

cvCamShift（backProject,searchWin,cvTermCriteria（CV_TERMCRIT_EPS|

CV_TERMCRIT_ITER,15,0.1）,&comp,&faceBox）;

searchWin=comp.rect;

}

SampleCodeTemplateMatching

 Collapse | CopyCode

//TemplateMatchingforEyedetection

voidFace:

:

findEyes_TM（IplImage*faceImage,TrackingSettings*settings）

{

CvSizefaceSize;faceSize=cvGetSize（faceImage）;

//LoadTemplatefromtheeyedatabase

CStringfileName;

//Nameofthetemplateforlefteye

fileName.Format（"%s\\eye%d.jpg",settings->params->DBdirectory,0）;

IplImage*eyeImage_Left=cvLoadImage（fileName,-1）;

//Nameofthetemplateforlefteye

fileName.Format（"%s\\eye%d.jpg",settings->params->DBdirectory,1）;

IplImage*eyeImage_Right=cvLoadImage（fileName,-1）;

IplImage*tempTemplateImg_Left;IplImage*tempTemplateImg_Right;

IplImage*templa