Diyabetik Retinopati tespitinde yeni bir algoritma kullanılarak optik disk yerinin kestirimi

Mehmet 1(5*ø=*1_{, ùH\KPXV$5,}2_{, Mehmet AKIN}3 1 Dicle Üniversitesi, Bilgisayar 0KHQGLVOL÷L%|OP'L\DUEDNÕU 2 Dicle Üniversitesi, *|]+DVWDOÕNODUÕ$QDELOLP'DOÕ,2'L\DUEDNÕU 3 'LFOHhQLYHUVLWHVL(OHNWULN(OHNWURQLN0KHQGLVOL÷L%|OP'L\DUEDNÕU

Özet

Bu oDOÕúPDGD, J|]GLELUHVLPOHUL]HULQGH J|UQWLúOHPHWHNQL÷LLOHDiyabetik Retinopati (DR) JLELELUWDNÕP VÕN UDVWODQDQ UHWLQD KDVWDOÕklarÕQÕQ tespitini hedefleyen ELUoRN oDOÕúPDQÕQ temel YH LON DGÕPÕQÕ ROXúWXUDQ Optik Disk (OD) \HULQLQWHVSLWLQLQJHUoHNOHúWLULOPHVLDPDoODQPÕúWÕU

*|] GLEL J|UQWOHULnin |QFHOLNOH ÕúÕN úLGGHWL NDQDOÕ .RQWUDVW 6ÕQÕUODPDOÕ Adaptif Histogram EúLWOHPHVLQH (KSAHE) WDEL WXWXOPXúWXU. Daha sonra Red Green Blue (RGB  UHQN X]D\ÕQGDNL J|UQWOHU gri skalaya G|QúWUOS EX J|UQWOHUH 0RUIRORMLN Kapama øúOHPL 0.ø X\JXODQPÕúWÕU Bu resimlere Canny Kenar Tespit (CKT) DOJRULWPDVÕ X\JXODQGÕNWDQVRQUDdHPEHUVHOHough D|QúP d+' DOJRULWPDVÕLOHEXOXQDQ bu kenarlar ]HULQGHki EHOLUOL ELU \DUÕoDS DUDOÕ÷ÕQD sahip oHPEHULPVi alanlar WHVSLW HGLOPLúWLU OD DODQÕQ VDKLS ROGX÷X VDUÕ UHQJL WHPVLO HGHQ \HúLO renk kanalÕQGDNL HúLN GH÷HUL EX oDOÕúPD LoLQ JHOLúWLULOPLú RODQ \LQHOHPHOL ELU DOJRULWPD LOH EHOLUOHQPLúWLU OD DGD\Õ oHPEHUOHU incelenen UHVLPOHULQ \HúLO renk NDQDOÕ ]HULQGH PDVNHOHQHUHN EX PDVNH DODQODUÕQGDQ GDKD |QFH EHOLUOHQPLú RODQ HúLN GH÷HUL NXOODQÕODUDN EX oDOÕúPD\D|]JQ LNLDGHW|]QLWHOLNoÕNDUÕOPÕúWÕU.

*|]GLELUHVLPOHULQGHWHVSLWHGLOHQOD DGD\ÕoHPEHUOHULn oÕNDUÕODQ|]QLWHOLNOHULQLdRN.DWPDQOÕ$OJÕOD\ÕFÕ d.$ WLSLQGHELU<DSD\6LQLU$÷ÕQD(YSA) uygulayarak, EXoHPEHUOHUOD olan veya olmayan úHNOLQGHbir VÕQÕIODQGÕUÕOPD \DSÕODUDN EDúDUÕ RUDQÕ bir oftalmolojistin LNL IDUNOÕ GH÷HUOHQGLUPH\H J|UH  87.50 ve % 95.00 NDGDUEXOXQPXúWXU

Anahtar Kelimeler: Optik Disk, .RQWUDVW 6ÕQÕUODPDOÕ $GDSWLI +LVWRJUDP (úLWOHPHVL, Morfolojik .DSDPDøúOHPL, dHPEHUVHO+RXJK'|QúP, dRN.DWPDQOÕ$OJÕOD\ÕFÕ

*_{<D]ÕúPDODUÕQ\DSÕODFD÷Õ\D]DU0HKPHW1(5*ø=PQHUJL]@dicle.edu.tr; Tel: (412) 248 80 30 (3673)}

Diyabetik Retinopati tespitinde yeni bir algoritma

NXOODQÕODUDNRSWLNGLVN\HUinin kestirimi

Cilt: 3-9

The estimation of optic disc location

via a novel algorithm for Diabetic

Retinopathy detection

Extended abstract

In most of the automated retinal image analysis systems Optic Disc(OD) localization is a main step. The position and region of OD is significantly important in terms of a few points. First of all, the location of macula can be detected using the location of OD. Since the fact that the blood vessels originate from the OD region in the embryonic period, the OD location is also used as the seed point for vessel extraction algorithms. More and more, the color range of the exudates, especially the hard ones, is so similar to the color range of the OD. Thus, the localization of the OD is a main step to be able to differentiate between the exudates and OD in this yellowish color range. In the color fundus images the OD can be observed as circle or ellipse like yellowish region where red blood vessels and optic nerves originate from inside of it. The other name of the OD is blind spot because of the fact that it contains no photoreceptor. In a normal fundus image the diameter of the OD is in the range of 80 and 100 pixels. The main disadvantages of OD localization are the inhomogeneous light distribution of the light in the color fundus images and the fact that the red blood vessels locate on this yellow region as well as extremely irregular OD shapes (Kaur and Sinha, 2012).

The DRIVE image database has been used for the evaluation of the implemented algorithms.In order to get rid of the inhomogeneous light distribution over the fundus images, the images are converted from Red Green Blue (RGB) color space to Hue Saturation Intensity (HSI) color space and then the intensity channel has been equalised using Contrast Limited Adaptive Histogram Equalization (CLAHE). After CLAHE algortihm has been applied the HSI color space has been converted back to the RGB color space. This new RGB image has been

converted to Grayscale format. The Grayscale image has been applied morphological closing operation with a disk structuring element of diameter 10. Afterwards, the Canny Edge Detection (CED) algorithm has been applied to the closed image with a threshold of 0.1 and the resulted edges has been applied Morphological Closing Operation (MCO) with a disk structruing element of a diameter value within 3 and 10. Finally, the Circular Hough Transform (CHT)algorithm has been applied over these edges and all circular patterns as an OD candidate has been localised.

Two problem specific features are extracted for each circle to be tested whether it is in OD area or not during the classification phase. One of these features is the multiplication of two extracted features. In order to get these extracted features, a threshold representing the yellowish region in green channel histogram is iteratively calculated by a novel algorithm. The first extracted feature is the ratio of the region whose pixel values are above this yellowish threshold to the whole masked region. The second extracted feature is the count of the pixels whose pixel values are above this yellowish threshold. The first feature is calculated as the multiply of these two extracted features. The other feature is a flag which is set as true only for the circle which has the maximum value of the first feature.

Each detected circle has been classified by applyinng its features to a Multi Layer Perceptron (MLP). The success ratio is 95.00 % for 20 training images and 20 testing images. This is a novel method for OD localization without contour detection which may be a basic step for the other retinal lesion detection systems.

Keywords: Optic Disc, Contrast Limited Adaptive Histogram Equalization, Morphological Closing Operation, Canny Edge Detection, Circular Hough Transform, Multi Layer Perceptron

*LULú

Biyomedikal g|UQW LúOHPH DODQÕ VRQ \ÕOODUGD VÕNOÕNODoDOÕúÕODQYH\DNÕQ]DPDQGDWPGQ\DGD X\JXODPDODUÕQÕQ oRN GDKD EHOLUJLQ ELU úHNLOGH J|]OHQHFH÷L EHNOHQHQ ELU DODQ ROPXúWXU %L\RPHGLNDO J|UQW LúOHPHQLQ |]HOOLNOH VRQ ]DPDQODUGD LOJLOHQGL÷L DODQODUGDQ ELUL GH J|] GLEL J|UQWOHULGLU *|] GLEL J|UQWOHULQLQ LúOHQPHVL NRQXVXQXQ GH÷HUOL ROPDVÕQÕQ VHEHEL PHYFXW GXUXPGD LQFHOHQPHVL JHUHNHQ J|UQW VD\ÕVÕQÕQ PLO\RQODU VHYL\HVLQGH ROPDVÕQD UD÷PHQEXJ|UQWOHULLQFHOH\HFHNJ|]GRNWRUX

VD\ÕVÕQÕQ \HWHUVL] NDOPDVÕGÕU Diyabetik

Retinopati (DR) GQ\DGDHQVÕNUDVWODQDQUHWLQD KDVWDOÕNODUÕQGDQ ELULGLU YH WHGDYL HGLOPHPHVL GXUXPXQGD N|UON LOH VRQXoODQÕU (Q D]  \ÕO NDGDU GL\DEHW KDVWDVÕ RODQODUÕQ  µL EX KDVWDOÕ÷DPDUX]NDOÕU'Q\D]HULQGH\ÕOÕ YHULOHULQH J|UH  PLO\RQ GL\DEHW KDVWDVÕ ROGX÷XWDKPLQHGLOPHNWHGLUYHEXUDNDPÕQ

\ÕOÕQGD  PLO\RQD oÕNDFD÷Õ WDKPLQ

edilmektedir (Poddar vd., 2011 5DNDPODUÕQ

E\NO÷QQ \DQÕQGD JHo WHúKLV YH WHGDYLQLQ N|UONOH VRQXoODQGÕ÷Õ JHUoH÷L EX DODQGDNL J|UQW LúOHPH oDOÕúPDODUÕQÕ GDKD |QHPOL YH GH÷HUOLNÕOPDNWDGÕU

Gerek '5¶nin tespitinde ve evrelenmesinde JHUHNVHGHGL÷HUVÕNUDVWODQDQJ|]KDVWDOÕNODUGDQ olan YDúD BD÷OÕ Makula Dejenerasyonu (YBMD) JLELKDVWDOÕNODUÕQWHVSLWLQGHOptik Disk (OD) yerinin tespiti temel bir aGÕPWHúNLOHGHU '5¶GH HNVGDODU LOHOD EHQ]HU UHQN DUDOÕ÷ÕQD WHNDEO HWWL÷L LoLQ |QFHOLNOHOD yerinin tespit

HGLOPHVL JHUHNLU %XQXQ GÕúÕQGD

mikroanevrizma, iQWUDUHWLQDO PLNURYDVNOHU anormallikler ve QHRYDVNODUL]DV\RQ gibi DR belirtileri de retinadaki dDPDU GD÷ÕOÕPÕ YH \|QHOLPL LOH LOJLOLGLU 5HWLQDGDNL GDPDUODU

HPEUL\RQLN G|QHPGH OD E|OJHVLQGHQ

EDúOD\DUDNUHWLQD\D\D\ÕOGÕNODUÕLoLQV|]NRQXVX GL÷HU GDPDUVDO EHOLUWLOHULQ EHOLUOHQPHVLQGH GH

OD \HULoR÷X]DPDQELUEDúODQJÕoQRNWDVÕWHúNLO HGHU $\UÕFD PDNXODQÕQ \HULOD¶LQ \HUL WHPHO DOÕQDUDNGDKDQHWWHVSLWHGLOHELOGL÷LLoLQYBMD KDVWDOÕ÷Õ LoLQ GH LON DúDPDGD WHVSLW HGLOPHVL gereken genelde OD µLQ UHWLQD J|UQWVQGHNL yeridir.

OD, ùHNLO 1¶GH GH J|]OHPOHQHELOGL÷L JLEL retina J|UQWOHULQGH\XYDUODNYH\DHOLSVH\DNÕQúHNOH VDKLS YH LoLQGHQ NÕUPÕ]Õ UHQNOL NDQ GDPDUODUÕ JHoHQ JHQHOGH VDUÕPVÕ UHQNWH RODQ ELU DODQGÕU 2SWLN VLQLUOHU EX DODQGDQ UHWLQD\D \D\ÕOPD\D EDúODUODU %X DODQÕQ LoLQGH KLo ÕúÕN UHVHSW|U ROPDGÕ÷Õ LoLQ EX DODQD N|U QRNWD GD denmektedir. NormaO ELU UHWLQD J|UQWVQGH

OD¶LQoDSÕ- 100 SLNVHOFLYDUÕQGDGÕU(Kaur ve Sinha, 2012).

*|]GLELJ|UQWOHULQGHGLNNDWHGLOPHVLJHUHNHQ SDUDPHWUHOHUGHQELULGHJ|UQWOHUGHNLG]HQVL] ÕúÕN GD÷ÕOÕPÕGÕU +DWWD oR÷X J|] GLEL J|UQWVQGH UHWLQD UHQJL NÕUPÕ]Õ tonlarda OD

\XPXUWD VDUÕVÕ UHQJLQGH LNHQ ED]Õ UHVLPOHUGH UHWLQD UHQJL JUL YH\D \HúLOH \DNÕQOD ise ya VROXN VDUÕ YH\D NHQDUODUÕ oRN EHOLUVL] úHNLOGH

J|UQWOHQPLú RODELOPHNWHGLU %DúND ELU

problem ise OD¶LQ NHQGLVLQLQ VDUÕ UHQJH VDKLS ROPDVÕQD UD÷PHQ RUWDVÕQGDQ NÕUPÕ]Õ UHQNOL

GDPDUODUÕQ JHoL\RU ROPDVÕGÕU OD ile ilgili

lezyona sahip insanlarda ise OD úHNOL oRN

G]HQVL] YH VWDQGDUWODUÕQ GÕúÕQGD ELU KDO alabilmektedir.

%X NRQXGD \DSÕODQ oDOÕúPDODU JHQHO RODUDN J|] GLEL UHVLPOHULQGH 2'¶LQ \HU WHVSLWL YH\D NHnar VHJPHQWDV\RQX LOH LOJLOHQPLúWLU 0RUDOHV YG  <HU WHVSLWL LOH HOGH HGLOHQ VRQXo 2'¶LQ PHUNH]LQL HQ GR÷UX úHNLOGH WD\LQ HWPH YH\D oHYUHVLQLPLQLPXPKDWDLOHNDSVD\DQELUoHPEHU YH\D HOLSV LOH VÕQÕUODPD \D GD 2' DODQÕQÕ

QRNWDVDO RODUDN EHOLUOHPH úeklindedir.

6HJPHQWDV\RQ LOH DPDoODQDQ LVH \HU WHVSLWLQLQ ELUDGÕP|WHVLQHJHoHUHN2'¶LQNHQDUODUÕQÕWDP RODUDN EHOLUOHPH\H oDOÕúPDNWÕU 2'¶LQ \HU WHVSLWLQL DPDoOD\DQ oDOÕúPDODUÕQ |QHPOL ELU

NÕVPÕ UHWLQD WDEDNDVÕQGDNL GDPDUODUÕQ

HPEUL\RQLNVDIKDGD2'E|OJHVLQGHQEDúOD\DUDN ELU VDUPDúÕN JLEL \D\ÕOGÕ÷Õ JHUoH÷LQL GLNNDWH DODUDN EX GDPDUODUÕQ KDNLP \|QHOLPOHULQL PRGHOOH\HQSDUDEROH÷ULOHULQLQNHVLúWL÷LQRNWD\Õ EXOPD\D oDOÕúPÕúWÕU )RUDFFKLD YG  Abr`amoff ve Niemeijer, 2006; Youssif vd.,

2008).Yer tespitLQLDPDoOD\DQEDúNDoDOÕúPDODU

LVH2'¶LQJ|]GLELUHVLPOHULQGHNLKHPHQKHPHQ KHU ]DPDQ HQ SDUODN E|OJH ROGX÷X YH \XYDUODN KDWODUD VDKLS ROGX÷X J|]OHPL LOH J|UQWQQ úLGGHW NDQDOÕQÕ YH 2'µLQ oHPEHUVHOOL÷LQL DQD

UHIHUDQV ND\QD÷Õ DOPÕúODUGÕU &KDLFKDQD YG  <DYX] YG  6HJPHQWDV\RQ DPDoOÕ oDOÕúPDODUGD LVH NHQDU WHVSLWL LoLQ 6QDNH DOJRULWPDODUÕ /LX YH &KHQ  ùDEORQ (úOHúWLUPH \|QWHPOHUL /DORQGH YG  YH :DWHUVKHG JLEL VHJPHQWDV\RQ DOJRULWPDODUÕ NXOODQÕOPÕúWÕU 0RUDOHV YG  %X oDOÕúPDODU GDKD GHWD\OÕ RODUDN DúD÷ÕGD EHOLUWLOPLúWLU

/DORQGH YH GL÷HUOHUL  UHWLQD

J|UQWOHULQLQ\HúLOUHQNNDQDOÕQÕ+DDU'DOJDFÕN '|QúP \|QWHPL\OH SLUDPLWVHO DQDOL]H WDEL WXWDUDN 2'¶LQ \HULQL WHVSLW HWWLNWHQ VRQUD +DXVGRUII WDEDQOÕ úDEORQ HúOHúWLUPH \|QWHPLQL kullanarak kenar segmentasyonunu JHUoHNOHúWLUPLúWLU %X oDOÕúPDGD  DGHW UHWLQD J|UQWV ]HULQGH 2' \HUL WHVSLWLQGH  EDúDUÕHOGHHGLOPLúWLU

)RUDFFKLD YH GL÷HUOHUL  UHWLQDGDNL GDPDU \DSÕODUÕQÕQ 2' ]HULQGHQ \D\ÕOGÕ÷Õ J|]OHPLQH GD\DOÕ RODUDN GDPDU \|QHOLPOHULQL JHRPHWULN RODUDN PRGHOOHPLúWLU %X PRGHO LNL DQD GDPDU \|QHOLP H÷ULVLQLQ NHVLúLP QRNWDVÕQÕ EXOPD\Õ DPDoODU %X oDOÕúPDGD 67$5( UHWLQD YHUL WDEDQÕQGDNL  WDQHVL VD÷OÕNOÕ RODQ YH  WDQHVL OH]\RQ LoHUHQ WRSODP  WDQH UHWLQD J|UQWVQGHQ  WDQHVLQGH 2'¶LQ \HUL WHVSLW HGLOPLúWLU

$EUCDPRII YH GL÷HUOHUL  UHWLQD J|UQWV]HULQGH JH]GLUGL÷L úDEORQXQ WHNDEO HWWL÷L DODQD DLW |]HOOLNOH GDPDU \|QHOLPL ELOJLOHULQL LoHUHQ WRSODP  DGHW |]HOOL÷L HOGH HGLS N11 VÕQÕIODQGÕUÕFÕVÕ LOH \LQHOHPHOL RODUDN VÕQÕIODQGÕUDUDN 2'¶H RODQ X]DNOÕ÷ÕQÕ WDKPLQ HWWLNWHQ VRQUD HQ GúN X]DNOÕ÷D VDKLS úDEORQD DLWDODQÕ2'RODUDNWD\LQHWPLúWLU%XoDOÕúPDGD +ROODQGD J|UQWOHPH SURJUDPODUÕQGDQ HOGH HGLOHQWDQHVLH÷LWLPLoLQWDQHVLLVHWHVt LoLQ NXOODQÕODQ YH  RUDQÕQGD OH]\RQ LoHUHQ UHWLQDJ|UQWOHULQGHQWDQHVLQGH2'¶LQ\HUL WHVSLWHGLOPLúWLU (%99.99).

<RXVVLI YH GL÷HUOHUL  UHWLQDGDNL WP GDPDU \DSÕVÕQÕ HOGH HGLS EX GDPDUODUÕ \|QHOLPOHULQH J|UH VHJPHQWH HWWLNWHQ VRQUD EX

damDUVHJPHQWOHULQLELUNDUúÕODúWÕUPDOÕILOWUHGHQ JHoLUHUHN J|UQWGHNL 2' DGD\Õ RODELOHFHN \HUOHUL EHOLUOHPLú YH GDKD VRQUD EX ILOWUHQLQ ER\XWXQX GH÷LúWLUHUHN HQ GúN KDWD RUDQÕQD VDKLS DGD\Õ 2' RODUDN WD\LQ HWPLúWLU %X oDOÕúPDGD 67$5( UHWLQD YHUL WDEDQÕQGaki OH]\RQ LoHUHQ YH LoHUPH\HQ WRSODP  DGHW UHWLQD J|UQWVQGHQ  WDQHVLQGH EDúDUÕOÕ ROXQPXúWXU '5,9( UHWLQD YHUL WDEDQÕQGDNL  DGHW UHWLQD J|UQWVQGH LVH WPQQWHVSLWLQGHEDúDUÕOÕROXQPXúWXU

Chaichana YH GL÷HUOHUL  UHWLQD

J|UQWVQH 6REHO ILOWUHVL X\JXOD\DUDN

J|UQWGHNL NHQDUODUÕ oÕNDUÕS EX NHQDUODU LoLQGHNL HQ oHPEHUVHO \DSÕ\Õ WHVSLW HWPHN LoLQ

dHPEHUVHO +RXJK '|QúPQ d+'

NXOODQPÕúWÕU %X oDOÕúPDGD '5,9( UHWLQD YHUL WDEDQÕQGDNL  DGHW UHWLQD J|UQWVQGHQ 

tanesinde OD \HU WHVSLWLQGH EDúDUÕOÕ

ROXQPXúWXU

<DYX] YH GL÷HUOHUL  UHWLQD

J|UQWOHULQGHNL HQ SDUODN DODQODUÕ YH d+' LOH EHOLUOHGLNOHUL oHPEHUVHO DODQODUÕ D\UÕ D\UÕ DGD\ODU úHNOLQGH DOÕS NDUúÕODúWÕUPDGD EXOXQDUDN

OD \HULQL WHVSLW HWPLúOHUGLU %X oDOÕúPada .DUDGHQL] 7HNQLN hQLYHUVLWHVL 7ÕS )DNOWHVL

2IWDOPRORML E|OPQGHQ DOÕQDQ UHWLQD

J|UQWOHULQGHEDúDUÕHOGHHGLOPLúWLU Liu ve Chen (2010), retinal floresan anjiyogram J|UQWOHULQL |QFHOLNOH PRUIRORMLN LúOHPOHUGHQ JHoLULS 2WVX \|QWHPLQL X\JXOD\DUDk OD yerini WHVSLW HWWLNWHQ VRQUD VHJPHQWDV\RQ DPDFÕ\OD GDPDUODUÕ WHVSLW HGLS J|UQWGHQ oÕNDUÕS (÷LP 9HNW|U $NÕúÕ *9) \|QWHPLQL NXOODQDUDN

ODµLQNHQDUODUÕQÕEHOLUOHPLúWLU%XoDOÕúPDGD DGHW J|UQW LoLQOD yerinin tespitinde %96.7,

OD kenar belirleme DúDPDVÕQGD LVH 

EDúDUÕHOGHHGLOPLúWLU

0RUDOHVYHGL÷HUOHULUHWLQDJ|UQWOHULQL |QFHOLNOH 7HPHO %LOHúHQ $QDOL]L 3&$ \|QWHPL\OH JUL VNDOD\D oHYLUGLNWHQ VRQUD

6WRNDVWLN :DWHUVKHG DOJRULWPDVÕ LOHOD yer

tespitini ve kenar segmentasyonunu geUoHNOHúWLUPLúWLU%XoDOÕúPDGD'5,216UHWLQD YHUL WDEDQÕQGDNL  UHWLQD J|UQWVQGH

 RUDQÕQGD OD yer tespiti ve kenar VHJPHQWDV\RQXEDúDUÕVÕVD÷ODQPÕúWÕU

%XoDOÕúPDGDLVH'5,9(UHWLQDYHULWDEDQÕQGDNL J|UQWOHU EDúWD ÕúÕN HúLWOHPH YH Canny Kenar Tespit (CKT) JLEL |Q LúOHPOHUGHQ JHoLULOLS d+' YH dRN .DWPDQOÕ $OJÕOD\ÕFÕ d.$) NXOODQÕODUDN2'\HUWHVSLWLDPDoODQPÕúWÕU

Görüntü materyali

%X oDOÕúPDGD '5,9( UHWLQD YHUL WDEDQÕQGDQ DOÕQDQ  WDQHVL H÷LWLP  WDQHVL GH WHVW LoLQ NXOODQÕODQ WRSODP  DGHW UHWLQD J|UQWV NXOODQÕOPÕúWÕU Staal vd., 2004; Niemeijer vd., 2004 '5,9( UHWLQD YHUL WDEDQÕ +ROODQGD¶GDNL bir DR J|UQWOHPH SURJUDPÕQGDQ DOÕQPÕúWÕU *|UQWOHPH SURJUDPÕ - \Dú DUDOÕ÷ÕQGDNL  DGHW GL\DEHW KDVWDVÕQÕ NDSVDPDNOD EHUDEHU 40 adet J|UQW UDVWJHOH VHoLOPLúWLU +HU ELU J|UQW -3(* IRUPDWÕQGD ROXS  GHUHFHOLN J|UQW DODQÕ )29 LOH Canon CR5 non-mydriatic 3CCD kamera NXOODQÕODUDN oHNLOPLúWLU dHNLOHQ LON J|UQWOHULQ ER\XWODUÕ [ ROXS KHU ELU UHVPLQ J|UQW DODQÕ NÕUSÕODUDN oDSÕ \DNODúÕN  SLNVHO RODQ ELU oHPEHUL NDSVD\DQ [ ER\XWODUÕQGD J|UQWOHUHOGHHGLOPLúWLr.

%X UHWLQD J|UQWOHUL GH÷LúLN ÕúÕN GD÷ÕOÕPODUÕQD VDKLS ROXS OH]\RQ LoHUHQ YH LoHUPH\HQ WLSOHUGHGLU5HVLPOHULQoHúLWOLOL÷LQHEDNÕOGÕ÷ÕQGD oDOÕúPDQÕQ GH÷LúLN J|UQW |UQHNOHUL ]HULQGH \DSÕOPÕúROGX÷XJ|UOHELOLUùHNLO¶GHEXUHWLQD YHUL WDEDQÕQGDQ oHúLWOL J|UQW |UQHNOHUL J|VWHULOPLúWLU

a )

b )

d ) ùHNLO1. '5,9(UHWLQDYHULWDEDQÕQGDQ|UQHN J|UQWOHUDOH]\RQLoHUPH\HQSDUODNJ|UQW EOH]\RQLoHUPH\HQPDWJ|UQWFOH]\RQ LoHUHQSDUODNJ|UQWGOH]\RQLoHUHQPDW J|UQW

Yöntem

*|] GLEL J|UQWOHULQGH ùHNLO ¶GH

J|UOHELOHFH÷L JLEL ÕúÕN GD÷ÕOÕPÕ oR÷X ]DPDQ WUGHú ROPDPDNWDGÕU g]HOOLNOHOD yer tespiti \DSÕOÕUNHQSDUODNOÕN \DGDGL÷HUELUGH\LúOHÕúÕN úLGGHWL GD÷ÕOÕPÕ UHIHUDQV DOÕQDQ GH÷LúNHQOHUGHQ ELULROGX÷XYH\DUHQNNHVNLQOL÷LQLHWNLOHGL÷LLoLQ EX NDQDOGDNL GD÷ÕOÕPÕQ |QFHOLNOH WUGHú KDOH JHWLULOPHVL JHUHNPHNWHGLU %X DPDoOD Red Green Blue (RGB) UHQN X]D\ÕQGDNL J|UQW |QFHOLNOH +6, UHQN X]D\ÕQD G|QúWUOGNWHQ VRQUD ÕúÕN úLGGHWL NDQDOÕQD .RQWUDVW 6ÕQÕUODPDOÕ Adaptif Histogram (úLWOHPHVL DOJRULWPDVÕ X\JXODQPÕúWÕU%XLúOHPGHQVRQUDJ|UQWWHNUDU

5*% UHQN X]D\ÕQD \HQL KDOL LOH

G|QúWUOPúWU (Park vd., 2007). ùHNLO ¶WH

ÕúÕN úLGGHWL HúLWOHPH LúOHPLQLQ HWNLOHUL

J|]OHPOHQHELOPHNWHGLU a ) b ) ,úÕNúLGGHWLHúLWOHPH MorfoORMLNøúOHP Kenar belirleme dHPEHUWHVSLWL (úLNGH÷HULEHOLUOHPH g]niteOLNoÕNDUPD 6ÕQÕIODQGÕUPD

c ) d ) ùHNLO3. ,úÕN úLGGHWLHúLWOHPHLúOHPL DHúLWOHPHLúOHPLQGHQ|QFHNLSDUODNJ|UQW EHúLWOHPHLúOHPLQGHQVRQUDNLSDUODNJ|UQW c) HúLWOHPHLúOHPLQGHQ|QFHNLPDWJ|UQW GHúLWOHPHLúOHPLQGHQVRQUDNLPDWJ|UQW 0RUIRORMLNLúOHP

,úÕN úLGGHWL HúLWOHPH LúOHPL WDPDPODQGÕNWDQ VRQUD5*%UHQNX]D\ÕQGDNLJ|UQWJULVNDOD\D G|QúWUOS  SLNVHO oDSÕQGD ELU GLVN úHNOLQGHNL \DSÕVDO HOHPDQ LOH Morfolojik .DSDPD øúOHPL 0.ø X\JXODQPÕúWÕU .DSDPD LúOHPL LOH DPDoODQDQ 2' ]HULQGHNL NÕUPÕ]Õ GDPDUODUÕQRUWDGDQNDOGÕUÕOPDVÕGÕU'L÷HUELUoRN oDOÕúPDGD 2' ]HULQGHNL GDPDUODUÕ RUWDGDQ NDOGÕUPDN LoLQ WP GDPDU VLVWHPL WHVSLW HGLOLS RULMLQDO UHVLPGHQ oÕNDUWÕOPÕúWÕU 2\VDNL EX oDOÕúPDGD JUL VNDOD ]HULQGH NDSDPD LúOHPL X\JXOD\DUDND\QÕDPDFDGDKDSUDWLNELUúHNLOGH XODúÕOPDNWDGÕU ùHNLO ¶WH X\JXODQDQ LúOHPOHULQ HWNLOHULJ|]OHPOHQHELOPHNWHGLU. a ) b ) c ) ùHNLO4. *ULVNDODYH0.øDHúLWOHPH LúOHPLQGHQVRQUDNL5*%UHQNX]D\ÕQGDNL J|UQW EJULVNDOD\DG|QúWUOPúJ|UQW c) 0.ø¶QGHQVRQUDNLJULVNDODJ|UQW

Kenar belirleme

CKT DOJRULWPDVÕQGD J|UQW\H |QFHOLNOH *DXVVLDQ ELU ILOWUH X\JXODQDUDN JUOWOHUL D]DOWÕOÕU 'DKD VRQUD J|UQWQQ JUL VNDOD GH÷LúLPOHULQL WHPHO DODUDN JUDG\DQÕ oÕNDUÕOÕU *|UQWQ JUDG\DQÕQGDQ HOGH HGLOHQ oL]JLOHU DoÕVDO \|QHOLPOHULQH J|UH VÕQÕIODQGÕUÕOÕU YH bu

oL]JLOHULQ GHYDPOÕN J|VWHUPH |]HOOLNOHUL

SDUDPHWULN RODUDN EHOLUOHQHELOHQ HúLN GH÷HUOHUL LOH WHVW HGLOLS NHQDU ROXS ROPDGÕNODUÕQD NDUDU verilir (Canny, 1986).

*|UQW\H 0.ø X\JXODQGÕNWDQ VRQUD

J|UQWQQNHQDUODUÕQÕQoÕNDUÕOPDVÕDPDoODQPÕú YH5REHUWV3UHZLWW6REHOYH&.7DOJRULWPDODUÕ GHQHQGLNWHQVRQUD&.7 DOJRULWPDVÕQÕQHúLN

VHYL\HVLQGH HQ L\L VRQXo YHUGL÷L

J|]OHPOHQPLúWLU .HQDUODUÕ HOGH HGLOHQ ED]Õ görüQWOHUGH 2' NHQDUODUÕQÕQ NRSXN oÕNWÕ÷Õ YH oHPEHU L]OHQLPL YHUPHGL÷L J|]OHQPLúWLU %X

DPDoOD VRQUDNL DúDPDODUGD oÕNDUÕODQ

|]QLWHOLNOHUL GH÷HUOHQGLUHQ ELU NRQWURO

PHNDQL]PDVÕ \DUGÕPÕ\OD WHVSLW HGLOHQ

oHPEHUOHUGHQ HQ D] ELULQLQ 2' DGD\Õ ROXS RODPD\DFD÷Õ WHVW HGLOLU (÷HU WHVSLW HGLOHQ oHPEHUOHULQ KLo ELULQLQ 2' LOH oDNÕúPDGÕ÷Õ WHVSLW HGLOLUVH EX VRUXQX JLGHUPHN LoLQ \LQHOHPHOL RODUDN 0.ø X\JXODQÕU (OGH HGLOHQ NHQDUODUD KHU \LQHOHPHGH ELU ELULP DUWWÕUÕOPDN ]HUH oDSÕ  ELULP RODQ GLVN úHNOLQGHNL ELU \DSÕVDO HOHPDQ LOH 0.ø X\JXODQÕU .RQWURO PHNDQL]PDVÕ WHVSLW HGLOHQ oHPEHUOHUGHQ HQ D] ELULQLQ 2' RODELOHFH÷LQL NDEXO HGHQH NDGDU \D GD \DSÕVDO HOHPDQÕQ oDSÕ  ELULP RODQD NDGDU EX\LQHOHPHLúOHPLWHNUDUHGLOHUHN2'¶LQNRSXN RODQNHQDUODUՁ]HULQGHNLoHPEHULPVL|UQWQQ |QSODQDoÕNPDVÕVD÷ODQÕUKenarlara yinelemeli RODUDN 0.ø X\JXODQPDVÕ \|QWHPL oDOÕúPD\D |]J ROXS oDOÕúÕODQ '5,9( UHWLQD YHUL EDQNDVÕQGDNL EHOLUVL] 2' NHQDU |UQWVQH VDKLS J|UQWOHUGHNL oHPEHULPVL |UQWOHULn ÕVNDODnmamaVÕ LoLQ JHOLúWLULOPLúWLU ùHNLO  D¶ GD&.7LúOHPLQLQHWNLVLùHNLOE¶GH ve ùHNLO  F¶GH LVH \LQHOHPHOL RODUDN X\JXODQDQ 0.ø LúOHPLQLQHWNLVLJ|]OHPOHQHELOPHNWHGLU

a )

b )

c )

ùHNLO D .HQDU EHOLUOHPH LúOHPLQGHQ VRQUDNL ELU J|UQW E .HQDU EHOLUOHPH LúOHPLQGHn sonUDNL 2' NHQDUODUÕ EHOLUVL] RODQ EDúND ELU J|UQW F 2' NHQDUODUÕ EHOLUVL] RODQ J|UQWQQ  SLNVHO oDSÕQGDNL ELULP \DSÕ LOH 0.ø X\JXODQGÕNWDQ VRQUDNL 2' NHQDUODUÕ belirgin hali

Çember tespiti

ÇHD ELUJ|UQW]HULQGHNLEHOLUOLELU\DUÕoDSD VDKLS oHPEHUVHO |UQWOHUL WHVSLW HGHU %X \|QWHPLOHJ|UQWX]D\ÕQGDQHOGHHGLOHQNHQDU |UQWOHUL]HULQGHNLKHUELUQRNWD\ÕV|]NRQXVX \DUÕoDSDVDKLSELUoHPEHULQPHUNH]LYDUVD\DUDN EX \HQL VDQDO oHPEHULQ ]HULQGHNL KHU ELU QRNWDQÕQ SDUDPHWUH X]D\ÕQGDNL GH÷HUL Eir DUWWÕUÕOÕU YH EX LúOHP J|UQW X]D\ÕQGDNL NHQDUODUÕQ ]HULQGHNL KHU ELU QRNWD LoLQ X\JXODQGÕNWDQ VRQUD SDUDPHWUH X]D\ÕQGDNL HQ \NVHN GH÷HUH VDKLS NRRUGLQDWODU J|UQW X]D\ÕQGDoHPEHULPVL|UQWOHULQPHUNH]LRODUDN EHOLUOHQPLúROXURizon vd., 2005).

Görüntüdeki kenarlar belirlendikten sonra, ÇHD uygulanarak 17- SLNVHO DUDOÕ÷ÕQGD \DUÕoDSD VDKLS RODQ oHPEHULPVL |UQWOHU WHVSLW HGLOHUHN HQ D] ELU oHPEHULQOD¶LQ ]HULQH RWXUGX÷X \D

da minimum hata ile OD¶L NDSVDGÕ÷Õ

J|]OHQPLúWLUùHNLO¶GDWHVSLWHGLOHQ oHPEHUOHU J|]OHPOHQHELOPHNWHGLU

a )

b )

c )

ùHNLO6. Görüntü üzerinde tespit edilen oHPEHUOHUDWHVSLWHGLOHQoHPEHUOHUEEDúDUÕOÕ olan 1 nolu çemberin OD üzerindeki gösterimi FEDúDUÕVÕ]RODQQROXoHPEHULQJ|VWHULPL

(úLNGH÷HULEHOLUOHPH

Kaur vH 6LQKD  VDUÕ UHQN HúLN GH÷HULQL EHOLUOH\HELOPHN LoLQ \LQHOemeli bir algoritma JHOLúWLUPLúWLU %XoDOÕúPDGDLVHWPUHVPLQ\HúLO UHQN NDQDO KLVWRJUDPÕ ]HULQGH VDUÕ UHQJH WHNDEOHGHQ\DNODúÕNHúLNVHYL\HVLVDUÕPVÕDODQ WP \HúLO UHQN NDQDO KLVWRJUDPÕQÕQ \DNODúÕN RODUDN \LUPLGH ELUL RODFDN úHNLOGH \LQHOHPHOL RODUDN |]JQ ELU \|QWHPOH WHVSLW HGLOLU ùHNLO µGH \LQHOHPHOL RODUDNOD¶L WHPVLO HGHQ VDUÕ UHQNOL DODQÕQ HúLN VHYL\HVLQL WHVSLW HWPH VUHFL

KLVWRJUDP ]HULQGH J|VWHULOPLúWLU (1

histogramdaki mDNVLPXPQRNWD\Õ(2 minimum

QRNWD\Õ E3 ve E4 DQOÕN RODUDN \LQHOHPH

VUHVLQGHNL HúLN GH÷HULQLQ EXOXQGX÷X QRNWD\Õ E5 LVH HúLN GH÷HULQLQ oDOÕúPDQÕQ JHUL NDODQÕQGD GD NXOODQÕODFDN RODQ HQ VRQ DOGÕ÷Õ GH÷HUL J|VWHUPHNWHGLU. Algoritma temel olarak DúD÷ÕGDNLDGÕPODUGDQROXúPDNWDGÕU 1. E5 = E2 - ( E2 - E1 ) / 10 2. Histogramda E5 µWHQ  µ\D NDGDUNL SLNVHO VD\ÕVÕQÕ WRSODPBSLNVHO GH÷LúNHQLQHDW(1) 3. WRSODPBSLNVHO GH÷LúNHQLQL

KLVWRJUDPGDNL WRSODP SLNVHO VD\ÕVÕQD E|OSRUDQGH÷LúNHQLQHDW

4. (÷HU RUDQ GH÷LúNHQL ¶WHQ NoN ¶WHQ E\N LVH HúLN VHYL\HVLQL (5 µH ata ve bitir.

5. (÷HU RUDQ GH÷LúNHQL ¶GDQ E\N LVH E5µL (2¶\H DW YH  DGÕPGDQ LWLEDUHQ G|QJ\HGHYDPHW ùHNLO7. <LQHOHPHOLRODUDNHúLNGH÷HUL belirleme VUHFLQLQ\HúLOUHQNNDQDOÕ histRJUDPՁ]HULQGH g]QLWHOLNoÕNDUma

7HVSLW HGLOHQ oHPEHUOHULQ PDVNH úHNOLQGH

J|UQWQQ GH÷LúLN UHQN NDQDOODUÕQD

X\JXODQDUDNELUoRN|]QLWHOLNoÕNDUÕOÕSOD olup ROPDPDGXUXPXQDQHNDGDUED÷ÕPOÕGH÷LúWLNOHUL J|]OHPOHQPLú YH EX ELUoRN |]HOOL÷LQ LoLQGHQ sadece DúD÷ÕGDNL EX oDOÕúPD\D |]JQ RODUDN WDVDUODQPÕú RODQ  DGHW |]QLWHOLN VHoLOHUHN

VÕQÕIODQGÕUPD VUHFLQGH JLUGL RODUDN

NXOODQÕOPÕúWÕU%XLNL|]QLWHOLNDúD÷ÕGDNLJLELGLU 1) %X|]QLWHOLNÕúÕNHúLWOHPHVLLúOHPLQGHQ JHoLULOPLú5*%UHQNX]D\ÕQGDNLJ|UQWQQ \HúLO UHQNNDQDOÕQGDQHOGHHGLOHQDúD÷ÕGDNL LNLDOW|]QLWHOL÷LQoDUSÕPÕLOHHOGHHGLOLU $úD÷ÕGDNLLNLDOW|]QLWHOL÷LQGH oÕNDUÕODELOPHVLLoLQVDUÕUHQJLWHPVLOHGHQ GDKD|QFHGHQKHVDSODQPÕúRODQHúLNGH÷HUL NXOODQÕOÕU a) +HUELUPDVNHGHEXHúLNVHYL\HVLQLQ VWQGHGH÷HUHVDKLSSLNVHOVD\ÕVÕQÕQ WPPDVNHQLQSLNVHOVD\ÕVÕQDRUDQÕHOGH edilir. b) +HUELUPDVNHGHEXHúLNVHYL\HVLQLQ VWQGHGH÷HUHVDKLSSLNVHOVD\ÕVÕHOGH edilir. 2) %X|]QLWHOLNELULQFL|]QLWHOLNGH÷HULHQ \NVHNRODQoHPEHULoLQGL÷HUoHPEHUOHU LoLQLVHRODUDNDWDQÕU 6ÕQÕIODQGÕUPD ÇKA, ELUGHQID]ODDOJÕOD\ÕFÕG÷PYHNDWPDQ LoHUHQ ELU <DSD\ 6LQLU $÷Õ (YSA) modelidir. +HUELUG÷PGR÷UXVDOROPD\DQELUHWNLQOHúPH IRQNVL\RQX LoHULU YH D÷ JHUL \D\ÕOÕP DOJRULWPDVÕQÕ NXOODQDUDN H÷LWLOLU /LQHHU RODUDN VÕQÕIODQGÕUÕOPDVÕ PPNQ ROPD\DQ YHULOHUL VÕQÕIODQGÕUPDGDVÕNOÕNODNXOODQÕOÕU

7HVSLW HGLOHQ KHU ELU oHPEHUH DLW \HúLO UHQN NDQDOÕQGDQ GHUOHQHQ  DGHW |]nitelik ELU YHNW|U úHNOLQGH d.$¶\D X\JXODQPÕúWÕU %X YSA¶GD JLUGLOHUKHUELUUHWLQDJ|UQWVQQLoLQGe tespit HGLOPLú KHU ELU oHPEHUGHQ HOGH HGLOPLú  DGHW |]nitelik LoHUHQELUYHNW|UoÕNWÕLVHEXoHPEHULQ

OD ROXS ROPDGÕ÷ÕQÕ \D GDOD ile ne kadar

NHVLúWL÷LQL J|VWHUHQ  YH  DUDVÕQGDNL GH÷HUOHUGLU%LULQFL|]niteOLNGH÷HUOHULQRUPDOL]H HGLOPLú ROPDVÕQD UD÷PHQ LNLQFL |]nitelik GH÷HUOHULQLQ WPQQ  YH  DUDOÕ÷ÕQGD ROGX÷X LoLQ QRUPDOL]DV\RQD LKWL\Do GX\XOPDPÕúWÕU. $÷ÕQ WUDQVIHU IRQNVL\RQODUÕ VÕUDVÕ\OD tansig ve logsig, H÷LWLPIRQNVL\RQXtraingdm (Momentum JHUL \D\ÕOPDOÕ H÷LP Gúú DOJRULWPDVÕ), |÷UHQPH fonksiyonu learngdm (Momentum D÷ÕUOÕNOÕ YH WDUDIOÕ |÷UHQPH H÷LP Gúú DOJRULWPDVÕ) |÷UHQPH RUDQÕ  G|QJ VD\ÕVÕ 10000 ve performans hedefi 0.001 olarak EHOLUOHQPLúWLU$÷GH÷LúLNDUDNDWPDQVD\ÕODUÕLOH WHVWHGLOPLútir.

Deneysel sRQXoODU YH performans

dH÷HUOHQGLUPHVL

ÇKA VÕQÕIODQGÕUÕFÕVÕQÕQ IDUNOÕ WRSRORMLOHUL LoLQ performans bilgileri Tablo 1 YH 7DEOR ¶GH 20 DGHWJ|UQWQQH÷LWLPDGHWJ|UQWQQLVH WHVW LoLQ NXOODQÕOGÕ÷Õ EX oDOÕúPDGD ÇKA¶GD  DGHWJLULúYHDGHWoÕNÕúEHOLUOHQPLúWLU%XQXQOD berDEHU RSWLPXP VD\ÕGDNL JL]OL NDWPDQ Q|URQ VD\ÕVÕ D÷ GH÷LúLN DUD NDWPDQ VD\ÕODUÕ LOH WHVW edildikten sonra, Tablo 1µGHNL VRQXoODU

gözetilerek 2,Tablo 2‘deki sonuçlar gözetilerek 6 olarak EHOLUOHQPLúWLU dDOÕúPDQÕQ SHUIRUPDQVÕ LVHLNLIDUNOÕEDNÕúDoÕVÕLOH HOHDOÕQPÕúWÕU7DEOR ¶GH VRQXoODUÕ J|VWHULOPLú RODQ ELULQFL EDNÕú DoÕVÕQD J|UH LVH WHVSLW HGLOHQ oHPEHULQOD’in ]HULQGH KHUKDQJL ELU QRNWD\Õ NDSVDPD RUDQÕ WHPHO DOÕQDUDN H÷LWLP JHUoHNOHúWLULOGLNWHQ VRQUD

WHVW DúDPDVÕQGD VDGHFHOD ile birebir oturan

çemEHUOHU GR÷UX NDEXO HGLOHUHN SHUIRUPDQV % RODUDN EHOLUOHQPLúWLU 7DEOR ¶GH VRQXoODUÕ J|VWHULOPLú RODQ GL÷HU EDNÕú DoÕVÕQD J|UH LVH WHVSLW HGLOHQ oHPEHUOHUGHQ ELUHELU

OD¶LQ VWQH GHQN JHOHQOHU GR÷UX JHUL\H

NDODQODU LVH \DQOÕú RODUDN GH÷HUOHQGLULOHUHN H÷LWLOPLú YH \LQH EX úHNLOGH WHVW HGLOHUHN performans %87.5RODUDNWHVSLWHGLOPLúWLU

Tablo 1. (÷LWLPDúDPDVÕQGDOD LOHNHVLúHQYH\D |UWúHQ oHPEHUOHU UHIHUDQV DOÕQDUDN ÇKA YSA PRGHOLQGH IDUNOÕ JL]OL NDWPDQ VD\ÕODUÕQGDNL SHUIRUPDQVVRQXoODUÕ d.$ Mimarisi Performans (%) RMS 2-1-1 90.0 0.3162 2-2-1 95.0 0.2236 2-6-1 82.5 0.4183 2-12-1 77.5 0.5000 2-20-1 85.0 0.3873 Tablo 2. (÷LWLP DúDPDVÕQGD VDGHFH 2' LOH

|UWúHQ oHPEHUOHU UHIHUDQV DOÕQDUDN ÇKA YSA PRGHOLQGH IDUNOÕ JL]OL NDWPDQ VD\ÕODUÕQGaki SHUIRUPDQVVRQXoODUÕ d.$ Mimarisi Performans (%) RMS 2-1-1 82.5 0.4183 2-2-1 77.5 0.4743 2-6-1 87.5 0.3536 2-12-1 85.0 0.3873 2-20-1 77.5 0.4743

Sonuçlar

%X oDOÕúPDGD '5,9( UHWLQD YHUL EDQNDVÕQGDQ

DOÕQDQ UHWLQD J|UQWOHULQGH OD yeri tespiti

DPDoODQPÕúWÕU dDOÕúPDGD NXOODQÕODQ '5,9( YHUL WDEDQÕ GL÷HU ED]Õ oDOÕúPDODUÕQGD GD

NXOODQÕOGÕ÷Õ WDUDIVÕ] ELU GH÷HUOHQGLUPH

\DSDELOPH\H LPNDQ VD÷ODGÕ÷Õ YH D\QÕ ]DPDQGD GH÷LúLN ÕúÕN WRQODUÕQGD oHNLP DoÕODUÕQGD YH OH]\RQOX |UQHNOHUL GH NDSVD\DQ J|UQWOerden ROXúWX÷X LoLQ WHUFLK HGLOPLúWLU *|UQWOHULQ GH÷LúLN ÕúÕN WRQODUÕQD VDKLS ROPDVÕ GH]DYDQWDMÕ $+7 NXOODQÕODUDN ÕúÕN HúLWOHPH \|QWHPL LOH PLQLPL]H HGLOPLúWLU 2'¶LQ J|UQW ]HULQGH GH÷LúLN SR]LV\RQODUGD YH ER\XWODUGD EXOXQPDVÕ VRUXQXGDd+'NXOODQÕODUDNYH\HúLOUHQNNDQDOÕ ]HULQGHNL VDUÕ UHQJL WHPVLO HGHQ YH DGDSWLI ELU úHNLOGH WHVSLW HGLOHQ HúLN GH÷HUL NXOODQÕODUDN oÕNDUÕODQ |]QLWHOLNOHU \DUGÕPÕ\OD DúÕOPD\D oDOÕúÕOPÕúWÕU +HU QH NDGDU '5,9( YHUL WDEDQÕQGDNL J|UQW ER\XWODUÕ 584x565 olsa da

algoriWPD GH÷LúLN J|UQW ER\XWODUÕQGD

oDOÕúDFDN úHNLOGH NRGODQPÕúWÕU .XOODQÕODQ \|QWHPOHU GH÷LúLN oDOÕúPDODU LOH RUWDN X\JXODPDODUDVDKLSROVDGDVDUÕUHQNHúLNGH÷HUL EXOPDYHEXHúLNGH÷HULQLNXOODQDUDNoHPEHUVHO PDVNHOHUGHQ oÕNDUGÕ÷Õ |]QLWHOLNOHU ED÷ODPÕQGD EXoDOÕúPDGDIDUNOÕ\|QWHPOHUL]OHQPLúROXSELU oftalmolojistLQLNLIDUNOÕGH÷HUOHQGLUPHNULWHULQH J|UH  YH % 95.00 EDúDUÕ RUDQODUÕQD sahiptir. øNLQFLGH÷HUOHQGLUPHNULWHULQHJ|UHHOGH edilen sonuç LOH OLWHUDWUGH \DSÕODQ GL÷HU oDOÕúPDODUÕQ performansODUÕ YH EX oDOÕúPDODUGD NXOODQÕODQ J|UQW VD\ÕODUÕ Tablo 3’teki gibi NDUúÕODúWÕUÕOPÕú olup EX oDOÕúPDQÕQ SHUIRUPDQV GH÷HULnin OLWHUDWUGHNL GL÷HU oDOÕúPDODUÕQ SHUIRUPDQV VRQXoODUÕQD NXOODQÕODQ J|UQW VD\ÕVÕQÕQ GD GL÷HU oDOÕúPDODUGD NXOODQÕODQ görüntü VD\ÕODUÕQD\DNÕQ ROGX÷XJ|]OHQPHNWHGLU.

Tablo 3. dDOÕúPDQÕQ sonucunda elde edilen

SHUIRUPDQV GH÷HULQLQ YH NXOODQÕODQ J|UQW VD\ÕVÕQÕQ OLWHUDWUGHNL GL÷HU oDOÕúPDODUda NXOODQÕODQ J|UQW VD\ÕODUÕ YH EX oDOÕúPDODUÕQ SHUIRUPDQVODUÕLOHNDUúÕODúWÕUÕOPDVÕ dDOÕúPD Görüntü _6D\ÕVÕ Performans _(%) Nergiz 40 95 Yavuz - 85 Lalonde 40 93 Chaichana 40 97.5 Liu ve Chen 60 96.7 Foracchia 81 97.5 Youssif 81 98.77 Morales 110 86.89 Abr`amoff 1000 99.99

Kaynaklar

Abr`amoff, M. D., Niemeijer M., (2006). The automatic detection of the optic disc location in retinal images using optic disc location regression2006 International Conference of the IEEE Engineering in Medicine and Biology Society (2006), vol. 1, pp. 4432-4435.

Canny. J.F., (1986). A computational approach to edge detection. IEEE Trans. on Pattern Analysis and Machine Intelligence, vol: 8(6), pp. 679-698, November 1986

Chaichana, T., Yoowattana, S., Sun, Z., Tangjitkusolmun, S., Sookpotharom, S., Sangworasil, M., (2008). Edge detection of the optic disc in retinal images based on identification of a round shape, Communications and Information Technologies, 2008. ISCIT 2008. International Symposium, pp. 670 –674

Foracchia, M., Grisan, E., Ruggeri, A., (2004). Detection of optic disc in retinal images by means of a geometrical model of vessel structure, Medical Imaging, IEEE Transactions on, vol. 23, no. 10, pp. 1189-95

Kaur, J., Sinha, H.P.,(2012). Automated localisation of optic disc and macula from fundus images, International Journal of Advanced Research in Computer Science and Software Engineering, vol: 2, no: 4

Lalonde, M., Beaulieu, M., Gagnon, L., (2001). Fast and robust optic disc detection using pyramidal

decomposition and hausdorff-based template matching, Medical Imaging, IEEE Transactions vol: 20, no: 11, pp. 1193 – 1200

Liu, S., Chen, J., (2010). Detection of the optic disc on retinal fluorescein angiograms, Journal of Medical and Biological Engineering, vol. 31(6), pp. 405-412

0RUDOHV 6 1DUDQMR 9 3HUH] ' 1DYHD $ AlFD×QL] 0 (2011). Automatic detection of optic disc based on PCA and Stochastic Watershed, Signal Processing Conference (EUSIPCO), 2012 Proceedings of the 20th European, pp. 2605 – 2609, 27-31 August 2012, Bucharest.

Niemeijer, M., Staal, J.J., van Ginneken, B., Loog, M., Abramoff, M.D., (2004). Comparative study of retinal vessel segmentation methods on a new publicly available database, SPIE Medical Imaging, vol.5370, pp. 648-656. Park, J., Kien, N.T., Gueesang, L., (2007). 2007

IEEE International Conference on Intelligent Computer Communication and Processing, pp. 237 - 241.

Poddar, S., Jha, B. K., Chakraborty, C., (2011). Quantitative clinical marker extraction from colour fundus images for non-proliferative diabetic retinopathy grading, International Conference on Image Information Processing (ICIIP 2011) , pp. 1 – 6

Rizon, M., Yazid, H., Saad, P., Shakaff, A. Y., Saad, A. R., Sugisaka, M., Yaacob, S., Mamat, M. R., Karthigayan, M., (2005).Object detection using circular hough transform, American Journal of Applied Sciences, vol: 2 (12), pp. 1606-1609 Staal, J.J., Abramoff, M.D., Niemeijer, M.,

Viergever, M.A., van Ginneken, B., (2004). Ridge based vessel segmentation in color images of the retina, IEEE Transactions on Medical Imaging, vol. 23, no: 4, pp. 501-509. Yavuz, Z., øNLEDú&ùHYLN8.|VH&A

method for automatic optic disc extraction in retinal fundus images, 5th International Advanced Technologies Symposium, pp. 93-98, May 2009, Karabuk, Turkey.

Youssif, A. A. A., Ghalwash, A. Z., Ghoneim, A. A. S. A., (2008). Optic disc detection from normalized digital fundus images by means of a vHVVHOV¶ direction matched filter, IEEE Trans Med Imaging., vol. 27(1), pp. 11-18, JANUARY 2008.