Segmentation of Historical Handwritten Documents into Text Zones ...

Report 2 Downloads 34 Views
2014 14th International Conference on Frontiers in Handwriting Recognition

Segmentation of Historical Handwritten Documents into Text Zones and Text Lines 

Basilis Gatos, Georgios Louloudis and Nikolaos Stamatopoulos Computational Intelligence Laboratory, Institute of Informatics and Telecommunications National Center for Scientific Research “Demokritos” GR-153 10 Agia Paraskevi, Athens, Greece {bgat, louloud, nstam}@iit.demokritos.gr   Abstract² In order to achieve accurate text recognition performance for historical handwritten document images, robust and efficient page segmentation is necessary. In this paper, we propose a text zone detection followed by a text line segmentation method suitable for historical handwritten documents. Our aim is to handle several challenging cases such as horizontal and vertical rule lines overlapping with the text, two column documents and characters of different text lines touching vertically. For text zone detection, we analyze vertical rule lines, connected components as well as vertical white runs while for text line segmentation, we enhance an existing approach based on Hough transform in order to better treat cases of vertical connected characters. Both methods have been proved very promising after an evaluation using a set of historical handwritten documents.

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±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

Keywords- historical document image processing; page segmentation; text line segmentation

,

,1752'8&7,21

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

2167-6445/14 $31.00 © 2014 IEEE DOI 10.1109/ICFHR.2014.84

464

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

465

WH[WOLQHDQGPDUNVWKHFRPSRQHQWVWKDWPDNHWKHOHWWHUVDQG ZRUGV RI LW $W D ILQDO VWHS WKH XQPDUNHG FRPSRQHQWV DUH DVVLJQHG WR WKH FORVHVW WH[W OLQH )RU WKH VHFRQG DOJRULWKP JUD\VFDOH FDVH  WKH VHDPV DUH FDOFXODWHG RQ WKH GLVWDQFH WUDQVIRUPRIWKHJUD\VFDOHLPDJH 0HWKRGV ZKLFK FDQQRW EH JURXSHG WR WKH DERYHPHQWLRQHG FDWHJRULHV LQFOXGH WKH ZRUN RI 6KL HW DO >@ ,Q WKLV ZRUN WKH DXWKRUV PDNH XVH RI WKH $GDSWLYH /RFDO &RQQHFWLYLW\ 0DS WKH LQSXW WR WKH PHWKRG LV D JUD\VFDOHLPDJHDQGDQHZLPDJHLVFDOFXODWHGE\VXPPLQJ WKH LQWHQVLWLHV RI HDFK SL[HO¶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etection of Vertical Text Zones based on Vertical Rule Lines 9HUWLFDO UXOH OLQHV DUH GHWHFWHG EDVHG RQ D IX]]\ VPRRWKLQJPHWKRG:HILUVWSURFHHGWRDYHUWLFDOVPRRWKLQJ XVLQJWKH5XQ/HQJWK6PRRWKLQJ$OJRULWKP 5/6$ >@LQ RUGHUWRFRQQHFWYHUWLFDOEURNHQUXOHOLQHV VHH)LJ $WD QH[W VWHS SL[HOV WKDW EHORQJ WR ORQJ YHUWLFDO OLQHV DUH GHWHFWHG E\ DSSO\LQJ WKH YHUWLFDO IX]]\ 5/6$ >@ 7KH IX]]\ 5/6$ PHDVXUH LV FDOFXODWHG IRU HYHU\ SL[HO RQ WKH

LLQLWLDO LPDJH DQG GHVFULEHHV ³KRZ IDU RQH FDQ VHH ZKHQ VVWDQGLQJDWDSSL[HODORQJYHUUWLFDOGLUHFWLRQQ´%\DSSO\LQQJWKLV P PHDVXUH D QHHZ JUD\ VFDOH LPDJH LV FUHDWHG ZKLFK LVV WKHQ EELQDUL]HG DQGG WKH YHUWLFDO UXOH OLQHV DUH H[WUDFWHG VHH )LJ  ,QRUGHUWRGHWHFWGRPLQDDQWYHUWLFDOUXOHOLQHVZHSUURFHHG WWRDYHUWLFDOSUURMHFWLRQRIWKKHUHVXOWLQJLP PDJH6LQFHZHHZDQW WWR VWUHVV WKH HH[LVWHQFH RI ORRQJ YHUWLFDO OLQHV ZH GHILQQH WKH YYHUWLFDOEODFNUXQSURILOHV xx DVIROORZV

SUHVHQWHGLQ)LJ$WDQH[W VWHSZHGHWHFFWDQGWDJLPDDJH ZKLWH UXQV ,I ZH FROXXPQV WKDW GRR QRW KDYH ORQQJ YHUWLFDO Z FDOFFXODWH DQG VRRUW DOO WKH OHHQJWKV RI VXXFFHVVLYH WDJJJHG FROXXPQVWKHQLIWWKHWZRORQJHUUOHQJWKVKDYHDOPRVWWKHVDPH YDOXXH WKH GRFXP PHQW LV FODVVLIILHG DV WZRFRROXPQ GRFXPHHQW 2WKKHUZLVHWKHP PDLQWH[WDUHD LVGHWHFWHG( ([DPSOHVRIERRWK FDVHVDUHSUHVHQWHHGLQ)LJ 

஻ሺ௫ሻሻ

ܸሺ‫ݔ‬ሻ ൌ ෍ ‫ܮ‬ሺ݅ǡ ‫ݔ‬ሻଶ 

 

௜ୀ଴

Z ZKHUHB(x) LV WKHQXPEHURRIEODFNUXQVLLQWKHFROXPQ xDQG L L(i,x) LV WKH OOHQJWK RI WKH iWK EODFN UXXQ $Q H[DPSSOH RI YYHUWLFDOSURMHFFWLRQVLVSUHVHQQWHGLQ)LJDD%\VHOHFWLQJJRQO\ WWKRVHLPDJHFFROXPQVWKDWKKDYHV(x) JUHDDWHUWKDQDWKUHHVKROG TZHFDQGHIL ILQHWKHYHUWLFDDOUXOHOLQHVHJJPHQWVRIWKH LPDJH VHH)LJE  )URPRXUH[SSHULPHQWVDVXXIILFLHQWYDOXHHIRUT LLV [ AH 2Z ZKHUHAH LVWKKHDYHUDJHFKKDUDFWHUKHLJKWW7KH P PDLQ WH[W ERGG\ FDQ EH GHIILQHG LQ WKH ]]RQH EHWZHHQ WKRVH FFRQVHTXHQW YHHUWLFDO UXOH OLQQHV WKDW KDYH PD[LPXP GLVVWDQFH EEHWZHHQWKHP P VHH)LJE 

DD  E 



E 



C. Horizontal Reestriction andd Refinement off Text Zones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iZLWKERXQGLLQJ ER[[ሺܿ‫ݔ‬ଵ௜ ǡ ܿ‫ݕ‬ଵ௜ െ  ܿ‫ݔ‬ଶ௜ ǡ ܿ‫ݕ‬ଶ௜ ሻEHORRQJVWRDYHUWLFFDOWH[W]RQHWKKDW LV GHILQHG EE\ [RIIVHWV x1 DQG x2 RQO\ LI ‫ݔ‬ଵ௜ ൑ ‫ݔ‬ଶ 9DOXXHVx1DQGx2DDUHWKHQXSGDWWHG ܿ‫ݔ‬ଶ௜ ൒ ‫ݔ‬ଵ ܿ‫ݔ‬ DFFFRUGLQJO\ EDVHHG RQ WKH OLLPLWV RI DOO YDOLG FRQQHFWWHG FRP PSRQHQWV,QRRUGHUWRKRUL]RRQWDOO\UHVWULFWWWKHYHUWLFDOWH[W ]RQQHV ZH FDOFXODWH WKH FRUUHVSRQGLQJ \RIIIVHWV y1 DQG y2 DV IROOORZV ‫ݕ‬ଵ ൌ ܿ‫ݕ‬ଵ௜ ™Š‡”‡‡ ݅ǣ ܿ‫ݕ‬ଶ௜ െ ܿ‫ݕ‬ଵ௜ ൒ ‫ݕܿܪܣ‬ ‫ݕ‬ଵ௜ ൌ ݉݅݊ 

) )LJXUH $QLP PDJHSRUWLRQEHIRUUH D DQGDIWHU E YHUWLFDOVPRRWKLQJ

D 

E 

)LJXXUH D 9HUWLFDOEODFNUXQSURMHHFWLRQVRILPDJH LQ)LJDQG E  WKH UHVXXOWLQJYHUWLFDOUXOHHOLQHVDVZHOODVWWKHGHWHFWHGPDLQQWH[W]RQH



D 





) )LJXUH 9HUWLFFDO IX]]\ 5/6$ D  RULJLQDO LPDJJH E  WKH IX]]\\ 5/6$ UUHVXOW

B B. Detection of Vertical Teext Zones baseed on Vertical White Runns 3URFHVVLQJJ WKH EDFNJURXXQG RI WKH LP PDJH FDQ EH XVVHG WR HHIILFLHQWO\ GHWWHFW WKH YHUWLFFDO WH[W ]RQHVV IRU WKH FDVHHV WKDW YYHUWLFDO UXOH OOLQHV GR QRW H[LVW ,Q RXU DSSURDFK ZHH ILUVW FFRXQW WKH YHHUWLFDO ZKLWH UXQV RI WKH LPDJH WKDW DDUH RI VVLJQLILFDQWOHQQJWK !LPDJHBBKHLJKW 6RP PHH[DPSOHVRRIWKLV SSURFHGXUH IRRU VLQJOH DQG WZRFROXP PQ GRFXPHQWVV DUH

‫ݕ‬ଶ ൌ ܿ‫ݕ‬ଶ௜ ™Š‡”‡‡ ݅ǣ ܿ‫ݕ‬ଶ௜ െ ܿ‫ݕ‬ଵ௜ ൒ ‫ݕܿܪܣ‬ ‫ݕ‬ଶ௜ ൌ ݉ܽ‫ݔ‬

 

$Q H[DPSOH RI WKH GHWHFFWLRQ RI PDLQQ WH[W UHJLRQ LV SUHVHQWHGLQ)LJ 

466

,9 9 7(;7/,1((6(*0(17$7,,21



D 



F 

E  

G 

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³6XXEVHW´³6XEEVHW ´DDQG³6XEVHW´³6XEVHW´FRUUHVSRQGVWRRWKHPDMRULW\\RI WKH QRUPDO FKDDUDFWHU FRPSRQHQWV 7KH PRWLYDWLRQ IRU ³6XXEVHW ´ GHILQQLWLRQ LV WR H[FFOXGH DFFHQWV DQG FRPSRQHHQWV WKDWWDUHODUJHLQKKHLJKWDQGDUHHOLNHO\WREHORRQJWRPRUHWKKDQ RQHH WH[W OLQH ³³6XEVHW ´ FFRQWDLQV DOO ODUJH FRQQHFWWHG FRP PSRQHQWV /DUUJH FRPSRQHQQWV DUH HLWKHU FDSLWDO OHWWHUV RU FKDDUDFWHUVIURPDDGMDFHQWWH[WOLLQHVWRXFKLQJ )LQDOO\³6XEEVHW ´ VKRXOG FRQWDLLQ FKDUDFWHUV DV D DFFHQWV SXXQFWXDWLRQ PDUUNV DQGGVPDOOFKDUDFWWHUV IRUGHWDLOOVVHH>@  7KH+RXJKWUUDQVIRUPPDSSSLQJVWHSLVUHVSRQVLEOHIRUWWKH GHWHHFWLRQ RI OLLQHV WKDW LQWWHUVHFW ZLWK WKH FRQQHFWWHG FRP PSRQHQWVRIHDDFKWH[WOLQH2 2QO\FRQQHFWHHGFRPSRQHQWVVRI ³6XXEVHW ´ DUH IH IHG WR WKH +RXXJK WUDQVIRUP 7KH QXPEHU U RI SRLQQWVSHUFRQQHFFWHGFRPSRQHQQWWKDWFRQWULEXXWHWRWKH+RXXJK GRP PDLQLVSURSRUUWLRQDOWRLWVZ ZLGWK>@$UUHVXOWRIWKLVVWWHS LVSSUHVHQWHGLQ)LLJ





) )LJXUH D  F 2ULJLQDOLPDJHV E  G WKHLUORQQJYHUWLFDOZKLWHUUXQV LQ UUHG DQGWKHGHWHFFWHGPDLQWH[W]RQQHV   

 )LJXUH 5HVXOWRIWKH+RX 5 XJK7UDQVIRUPPDDSSLQJVWHS

D 



7KH SRVWSURFFHVVLQJ VWHS FFRQVLVWV RI WZ ZR VWDJHV $W WWKH ILUVVWVWDJHDPHUJJLQJWHFKQLTXHHRYHUWKHUHVXXOWRIWKH+RXXJK WUDQQVIRUP LV DSSSOLHG WR FRUUHHFW SRVVLEOH IDOVH I DODUPV DDQG FRQQQHFWHGFRPSRRQHQWVRI³6XEEVHW´WKDWZ ZHUHQRWFOXVWHUUHG WRDDQ\WH[WOLQHDUUHFKHFNHGWRVHHZKHWKHUWKKH\FUHDWHDQQHZ WH[WWOLQHWKDWWKH+ +RXJKWUDQVIRRUPGLGQRWUHYYHDO 7KHVHFRQGSSRVWSURFHVVLQJJVWDJHGHDOVZ ZLWKFRPSRQHHQWV O\LQQJLQ³6XEVHW ´7KLVVXEVHHWLQFOXGHVFRP PSRQHQWVZKRRVH KHLJJKW H[FHHGV WKUHH WLPHV WKH DYHUDJH KHLJKW AH $ $OO

E 

)LJXUH D %LQDUL]HGLPDJJH E WKHGHWHFWHHGPDLQWH[WUHJLRRQ

467

FFRPSRQHQWV RRI WKLV VXEVHW PDLQO\ EHORQJJ WR n GHWHFWHHG WH[W 2XUQRYHOPHWKKRGIRUVSOLWWLQQJWKHVHFRPSRRQHQWV OOLQHV n! 2 FFRQVLVWVRIWKHHIROORZLQJVWHHSV $  &DOFXXODWH yi ZKLFKK DUH WKH DYHUUDJH y YDOXHV RI WKH LLQWHUVHFWLRQRIIGHWHFWHGOLQHiDQGWKHFRQQQHFWHGFRPSRQQHQW¶V EERXQGLQJER[ i «n  VHHH)LJDI  PSXWH WKH VNHOHHWRQ RI WKH FRRQQHFWHG FRPSSRQHQW %  &RPS )LJ E J  GHWHFW DOO MXXQFWLRQ SRLQWVV DQG UHPRYH WKHP IIURPWKHVNHOHHWRQ )LJF K ,IQRMXQFFWLRQSRLQWH[[LVWVLQ ZHHQ \L DQG \L WWKH ]RQH EHWZ L  UHPRYH DOOO VNHOHWRQ SRLLQWV LQ WWKHFHQWHURIWKKH]RQH &  $VVLJJQ WKH VNHOHWRRQ FRQQHFWHG FRPSRQHQWV WWR WKH FFORVHVWOLQH\L )LJGL  '  (DFKKSL[HORIWKHLLQLWLDOLPDJH )LJDI LQQKHULWV WWKHWH[WOLQHLGGRIWKHFORVHVWWVNHOHWRQSL[HHO )LJHM 



D 

E 



L 



H 

J 







K 

M 



)LJXUH 6WHSV IRU DVVLJQLQJ SL[[HOV RI YHUWLFDOO\ FRQQHFWHG FKDUDFWHUV WR ) FFRUUHVSRQGLQJWH[[WOLQHV FRORUHGOOLQHV  DI ,QLWLDDOLPDJH EJ VNNHOHWRQ F K  UHPRYDO RII MXQFWLRQ SRLQWV G L  DVVLJQPHQQW RI VNHOHWRQ FRQQQHFWHG FFRPSRQHQWVWRFORRVHVWOLQH HM ILLQDOUHVXOW

9



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i WKHVHWRIDOOSRLQQWVRIWKHiJURXXQGWUXWKWH[WOLQHRjWKHVHWWRI DOO SRLQWVRIWKHjjWH[WOLQHUHVXXOW(s)DIXQFWLRQWKDWFRXXQWV MatchScore(i,jj) UHSUHVHQWV WWKH WKH HOHPHQWV RI VHW V 7DEOH M PDWWFKLQJUHVXOWV RIiJURXQGWUUXWKWH[WOLQHDDQGWKHjWH[WOLQH UHVXXOWDVIROORZV T Gi ˆ R j   MatchhScore i j  T Gi ‰ R j :HFRQVLGHU DUHJLRQSDLU DVDRQHWRRQHPDWFKRQO\ \LI WKH PDWFKLQJ VFFRUH LV HTXDO WR RU DERYHH WKH HYDOXDWRRU V N LV WKH FRXQW RI JURXQGWUXXWK DFFFHSWDQFH WKUHVKKROG 7D ,I N HOHP PHQWV M LV WKKH FRXQW RI UHHVXOW HOHPHQWVV DQG o2o LV WWKH QXP PEHU RI RQHWWRRQH PDWFKHHV ZH FDOFXODDWH WKH GHWHFWLLRQ UDWHH DR DQGUHFFRJQLWLRQDFFXXUDF\ RA DVIIROORZV



F 



G 

I 





)LJXUH 7KHJURXQGWUXWKR 7 RIPDLQGRFXPHQWWWH[W]RQHV

oo oo  RA    N M  7KH )0HDVXXUH PHWULF F FM  FDQ EH H[WUDFWHG LI ZH FRP PELQH WKH YDDOXHV RI GHWHHFWLRQ UDWH DQG D UHFRJQLWLLRQ DFFFXUDF\  DR RA  F FM   DR  RA DR R

(;3(5 5,0(17$/5(6 68/76

,Q RUGHU WR UHFRUG WKH HIILFLHQF\ RI WKH GHYHORSHG WH[W ZULWWHQ ]]RQHGHWHFWLRQQ PHWKRGZH FUHDWHGDVHW RIKDQGZ KKLVWRULFDO LPDDJHV IURP WKH WUDQ6FULSWRULXXP SURMHFW >@ WKDW KKDYH UHSUHVHQQWDWLYH OD\RXW FKDOOHQJHV RRQH RU WZR FRROXPQ GGRFXPHQWV VLLGH QRWHV KRRUL]RQWDO DQG YHUWLFDO UXOH OLQHV ZHPDQXDOO\P PDUNHGWKHPDLLQWH[W VVHDOVHWF $WDQH[WVWHSZ ]]RQHV 7KHVH FRUUHVSRQG WR RQH DUHD IRU VLQJOH FRROXPQ GGRFXPHQWVRU WZRDUHDVIRU WZRFROXPQGGRFXPHQWV VHHH)LJ  

7KH SURSRVHGG WH[W OLQH VHJJPHQWDWLRQ KDDV EHHQ WHVWHG RQ  KDQGZULWWHQQ GRFXPHQW LPDJHV GHUULYHG IURP WWKH WUDQQ6FULSWRULXPSSURMHFW>@)RRUDOOWKHVHGRRFXPHQWLPDJJHV

468

ZH KDYH PDQXDOO\ FUHDWHG WKH FRUUHVSRQGLQJ WH[W OLQH VHJPHQWDWLRQ JURXQG WUXWK XVLQJ WKH $OHWKHLD WRRO >@ DQG VWRUHG LW XVLQJ WKH 3$*( [PO IRUPDW 7KH WRWDO QXPEHU RI WH[W OLQHV DSSHDULQJ RQ WKRVH LPDJHV ZDV  7KH DFFHSWDQFH WKUHVKROG IRU WKH WH[W OLQH VHJPHQWDWLRQ HYDOXDWLRQ LV VHW WR 7D   )RU WKH VDNH RI FRPSDULVRQ ZH DOVR WHVWHG WKH ZLQQLQJ PHWKRG RI WKH ,&'$5 +DQGZULWLQJ6HJPHQWDWLRQ&RPSHWLWLRQ>@IRUWKHWH[WOLQH VHJPHQWDWLRQ WDVN DV ZHOO DV WKH PHWKRG SURSRVHG E\ 1LFRODRXHWDO>@7DEOH,VKRZVFRPSDUDWLYHH[SHULPHQWDO UHVXOWVLQWHUPVRIGHWHFWLRQUDWH DR UHFRJQLWLRQDFFXUDF\ RA  DQG )0HDVXUH FM  $V LW FDQ EH REVHUYHG IURP 7DEOH , WKH SURSRVHG PHWKRG RXWSHUIRUPV WKH RWKHU WZR DSSURDFKHV DFKLHYLQJ GHWHFWLRQ UDWH  UHFRJQLWLRQ DFFXUDF\DQGFM 7$%/(, 0HWKRG

N

'RFXPHQW,PDJH$QDO\VLVIRU/LEUDULHVSS >@ ( %UX]]RQH DQG 0 & &RIIHWWL

$Q $OJRULWKP IRU ([WUDFWLQJ &XUVLYH 7H[W /LQHV

 WK ,QW &RQI RQ 'RFXPHQW $QDO\VLV DQG 5HFRJQLWLRQSS >@ 0 $ULYD]KDJDQ + 6ULQLYDVDQ DQG 6 1 6ULKDUL $ 6WDWLVWLFDO $SSURDFK WR +DQGZULWWHQ /LQH 6HJPHQWDWLRQ 'RFXPHQW 5HFRJQLWLRQDQG5HWULHYDO;,93URFRI63,(SS7 >@ / /LNIRUPDQ6XOHP $ +DQLP\DQ DQG & )DXUH

$ +RXJK %DVHG $OJRULWKPIRU([WUDFWLQJ7H[W/LQHVLQ+DQGZULWWHQ'RFXPHQWV

UG ,QW &RQI RQ 'RFXPHQW $QDO\VLV DQG 5HFRJQLWLRQ  SS   >@ */RXORXGLV%*DWRV,3UDWLNDNLV&+DODWVLV7H[WOLQHDQGZRUG VHJPHQWDWLRQ RIKDQGZULWWHQGRFXPHQWV 3DWWHUQ 5HFRJQLWLRQ 9RO ,VVXHSS >@ @ $1LFRODRXDQG%*DWRV+DQGZULWWHQ7H[W/LQH6HJPHQWDWLRQE\ 6KUHGGLQJ7H[WLQWRLWV/LQHVWK,QW&RQIRQ'RFXPHQW$QDO\VLV DQG5HFRJQLWLRQSS >@ 5 6DDEQL $ $VL - ( 6DQD

7H[W OLQH H[WUDFWLRQ IRU KLVWRULFDO GRFXPHQW LPDJHV

 3DWWHUQ 5HFRJQLWLRQ /HWWHUV 9RO  ,VV  SS  >@ = 6KL 6 6HWOXU DQG 9 *RYLQGDUDMX 7H[W ([WUDFWLRQ IURP *UD\ 6FDOH +LVWRULFDO 'RFXPHQW ,PDJHV 8VLQJ $GDSWLYH /RFDO &RQQHFWLYLW\ 0DS WK ,QW &RQI RQ 'RFXPHQW $QDO\VLV DQG 5HFRJQLWLRQSS >@ @ / /LNIRUPDQ  6XOHP $ =DKRXU DQG % 7DFRQHW

7H[W /LQH 6HJPHQWDWLRQ RI +LVWRULFDO 'RFXPHQWV $ 6XUYH\

 ,QWHUQDWLRQDO -RXUQDO RQ 'RFXPHQW $QDO\VLV DQG 5HFRJQLWLRQ ,-'$5  YRO  QRSS$SULO >@ % *DWRV , 3UDWLNDNLV DQG 6- 3HUDQWRQLV

$GDSWLYH GHJUDGHG GRFXPHQWLPDJHELQDUL]DWLRQ

3DWWHUQ5HFRJQLWLRQYROQRSS ±0DU >@ )0 :DKO .@ 1 6WDPDWRSRXORV * /RXORXGLV % *DWRV 8 3DO DQG $ $ODHL ,&'$5 +DQGZULWLQJ 6HJPHQWDWLRQ &RQWHVW WK ,QW &RQIHUHQFHRQ'RFXPHQW$QDO\VLVDQG5HFRJQLWLRQSS  >@ &&ODXVQHU63OHWVFKDFKHUDQG$$QWRQDFRSRXORV$OHWKHLD$Q $GYDQFHG 'RFXPHQW /D\RXW DQG 7H[W *URXQG7UXWKLQJ 6\VWHP IRU 3URGXFWLRQ (QYLURQPHQWV WK ,QW &RQI RQ 'RFXPHQW $QDO\VLV DQG5HFRJQLWLRQSS

&203$5$7,9((9$/8$7,215(68/76 M

o2o

DR  





RA   FM  

3URSRVHG

 





:LQQHU>@

 









1LFRODRX>@  









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aLVUHFRUGHGIRUERWKFDVHV $&.12:/('*0(17 7KHUHVHDUFKOHDGLQJWRWKHVHUHVXOWVKDVUHFHLYHGIXQGLQJ IURPWKH(XURSHDQ8QLRQ V6HYHQWK)UDPHZRUN3URJUDPPH )3  XQGHU JUDQW DJUHHPHQW Qƒ   WUDQ6FULSWRULXP 5()(5(1&(6 >@ KWWSZZZWUDQVFULSWRULXPHX >@ ) 6KDIDLW ' .H\VHUV DQG 7 %UHXHO

3HUIRUPDQFH HYDOXDWLRQ DQG EHQFKPDUNLQJ RI VL[SDJH VHJPHQWDWLRQ DOJRULWKPV

 ,((( 7UDQV 3DWW$QDO0DFK,QWHOO9ROQRSS-XQH >@ 90DOOHURQ9(JOLQ+(PSWR]6'RUG&URXVOH35HJQLHU

7H[W /LQHV DQG 6QLSSHWV ([WUDFWLRQ IRU WK &HQWXU\ +DQGZULWLQJ 'RFXPHQWV/D\RXW$QDO\VLV

WK,QW&RQIRQ'RFXPHQW$QDO\VLV DQG5HFRJQLWLRQSS >@ 0 %XODFX 5 YDQ .RHUW /6FKRPDNHU 7 YDQ GHU =DQW /D\RXW DQDO\VLVRIKDQGZULWWHQKLVWRULFDOGRFXPHQWVIRUVHDUFKLQJWKHDUFKLYH RI WKH &DELQHW RI WKH 'XWFK 4XHHQ WK ,QW &RQI RQ 'RFXPHQW $QDO\VLVDQG5HFRJQLWLRQSS >@ 66%XNKDUL70%UHXHO$$VL-(O6DQD³/D\RXW$QDO\VLVIRU $UDELF+LVWRULFDO 'RFXPHQW,PDJHV 8VLQJ0DFKLQH /HDUQLQJ´WK ,QW &RQI RQ )URQWLHUV LQ +DQGZULWLQJ 5HFRJQLWLRQ  SS   >@ 6 1LFRODV 7 3DTXHW DQG / +HXWWH &RPSOH[ +DQGZULWWHQ 3DJH 6HJPHQWDWLRQ 8VLQJ &RQWH[WXDO 0RGHOV³ QG ,QW :RUNVKRS RQ

469