ࡱ> .0-J7 w0bjbjUU H7|7|w+l , $e ; ; ; ?2$ ;!; 7 ; ; ;  ;;   ; V  `ﮅ  50eSpatio-temporal image segmentation using optical flow and clustering algorithm Saaa Gali and Sven Lon ari* Ericsson Nikola Tesla Krapinska 45, 10000 Zagreb, Croatia *Faculty of Electrical Engineering and Computing, University of Zagreb Unska 3, 10000 Zagreb, Croatia e-mail: Sasa.Galic@etk.ericsson.se, Sven.Loncaric@fer.hr Abstract Image segmentation is an important and challenging problem in image analysis. Segmentation of moving objects in image sequences is even more difficult and computationally expensive. In this work we propose a technique for spatio-temporal segmentation of medical image sequences based on clustering in the feature vector space. The motivation for spatio-temporal approach is the fact that motion is a useful clue for object segmentation. Two- dimensional feature vector has been used for clustering in the feature space. The first feature is image brightness which reveals the structure of interest in the image. The second feature is the Euclidean norm of the optical flow vector. The optical flow field is computed using a Horn-Schunck algorithm. By clustering in the feature space, it is possible to detect a moving object in the image. Experiments have been conducted using a sequence of ECG-gated magnetic resonance (MR) images of a beating heart. The method is also tested on images with moving background. The experiments have shown encouraging results. Keywords: spatio-temporal image segmentation, clustering, optical flow, image analysis 1. Introduction Segmentation is a challenging field of image analysis. In particular, medical image segmentation has become very important with development of complex medical imaging modalities which are capable of producing a large quantity of high-resolution two-dimensional (2-D) and three-dimensional (3-D) images. The problem of image segmentation has been studied extensively and there is a large number of methods described in the literature [1]. In many applications including medical image analysis it is necessary to analyze an image sequence of the same scene but at different time moments. In such a case, the obtained images often consist of a static background and moving object(s) of interest. An example of such a sequence is ECG-gated magnetic resonance (MR) image of the heart taken at different time moments. For example, in ECG-gated MR scanning one may acquire sixteen 3-D MR volumes during a single heart beat cycle. Manual analysis of such images by physicians is difficult and time consuming. Computer assisted methods are capable of faster and more accurate quantitative measurements required for medical treatment. In this work we propose a new technique for spatio-temporal segmentation based on clustering [3] in the feature vector space. Motion is a useful clue for object segmentation and that is the main motivation for the proposed spatio-temporal segmentation method. Computed features include image brightness and optical flow vector. The paper is organized as follows. In Section 2 a short overview of optical flow algorithm is presented. In Section 3 segmentation using clustering method is presented. Experimental results are shown in Section 4, and conclusion is given in Section 5. 2. Optical flow field computation In this work the algorithm proposed by Horn and Schunck [2] has been used. The algorithm determines the optical flow as a solution of the following partial differential equation:  EMBED Equation.3  (1) The solution of Equation 1 is obtained by numerical procedure for error function minimization. The error function E is defined in terms of spatial and time gradients of optical flow vector field and consist of two terms shown in Equations 2a-2e.  EMBED Equation.3  (2a)  EMBED Equation.3  (2b)  EMBED Equation.3  (2c)  EMBED Equation.3   EMBED Equation.3   EMBED Equation.3  (2d)  EMBED Equation.3   EMBED Equation.3  (2e) E (x, y, t) represents luminance of the image in point (x, y) at time moment t. To solve the minimization problem a steepest descent method is used which is based on computation of gradient to determine the direction of search for the minimum. The optical flow algorithm has two main phases. In the first phase, gradient coefficients Ex, Ey, Et are computed from input images. The coefficients represent estimates of the image gradients in space and time, and they are defined by Equation 2d. In the second phase, the optical flow vectors u and v defined by Equation 2e are computed. Depending on image content, the number of iteration steps must be chosen. If the moving object in the image has a big uniform area, the number of iterations must be large. If the image consists of small moving objects, the number of steps can be small. 3. Clustering-based segmentation Most of the classical image segmentation techniques rely only on a single frame to segment the image. However, motion is a very useful clue for image segmentation. The main idea of this work is to develop a spatio-temporal image segmentation technique for image sequences. In this approach segmentation is not done on a simple frame-by-frame basis but utilizes multiple image frames to segment the objects of interest. For this purpose we extract features both from the actual image that has to be segmented and from neighboring image frames in the sequence. The extracted feature vectors are clustered using a clustering algorithm to determine the characteristic image regions. The research of various feature vectors is underway and here we present the currently used features. The first feature is image brightness which is useful for segmentation because the heart regions of interest are bright while the background is mostly dark. The second feature is the Euclidean norm of the optical flow vector defined by Equation 4.  EMBED Equation.3  (4) By using the above features, we obtain both the spatial and temporal information about the scene. K-means clustering algorithm has been used in this work [3]. The feature space is divided into four characteristic areas corresponding with four image regions. The first region is the static background and the initial cluster center vector u1(0) for first group is set to (0,0). The second image region represents the moving background and has the initial cluster center vector u2(0) equal to (0,M), where M is maximum norm in Eof matrix computed by Equation 4. The third image region represents static objects with the initial cluster center vector u3(0) equal to (0,m) where m is the maximum image brightness. And finally the fourth region represents moving objects and has the initial cluster center vector u4(0) equal to (m, M). The rule for partitioning the feature space is defined by Equation 5.  EMBED Equation.3  j=1...K (5) where Rk is k-th cluster, K is the number of clusters (in our case K=4), and Uk(n) is the center of the k-th cluster. The clustering algorithm works iteratively where in each cluster the membership for each feature vector is determined and then new cluster centers are computed by Equation (6).  EMBED Equation.3  k=1...K (6) The process is repeated until centers stabilize, i.e. until uk (n+1) becomes equal to uk(n). The resulting clusters correspond to four characteristic image regions in the segmented image. Computed results must be taken with reserve because it is possible that some pixels are mistakenly classified to cluster three instead of four. This can happen if there is a very large motion in picture. With such a motion the object will be separated in two regions. The first region shows us the portion of the object that has very large motion, and the second region corresponds to the rest of the object. It is also important to mention that before the computation takes place, input features (image brightness and optical flow energy) are normalized to the same range of values (in our case we normalized optical flow energy according to brightness). 4. Experimental results Experiments have been conducted using ECG-gated MR heart images acquired at sixteen time moments. The procedure consists of two steps. In the first step the optical flow is computed for two input images with the same content but acquired at different time moments. The result, which represents the optical flow field, is saved in a file that is loaded by the program for clustering analysis that is performed in the second step. The result consists of four binary images representing the four characteristic regions of the input image. The results are shown in Figures 1 to 3. Figure 4 represents the segmented image computed by simple thresholding, and Figure 5 represents the segmented image computed using only the optical flow. Figure 6 shows the position of feature vector in the feature space with centers marked as boxes.   Figure 1 Input MR image. Figure 2 Optical flow field.   Figure 3 Segmented image using Figure 4 Segmented image using clustering method. threshold method.   Figure 5 Segmented image using Figure 6 Clusters and centers. optical flow us input parameter. The proposed method works correctly even in the case of moving background. In this case, the optical flow is shown in Figure7, while Figure 8 represents segmented image, which gives the same result as in the above experiment with mostly static background.   Figure 7 Computed optical flow. Figure 8 Segmented image using clustering method. 5. Discussion and conclusions In this work we have presented a method for spatio-temporal image segmentation which is based on clustering and optical flow computation. The experiments have been conducted using MR image sequence of the beating heart and have demonstrated the feasibility of the method. We have compared the results computed by clustering method with the results obtained by simple threshold segmentation method and with the results computed by applying optical flow vector as condition for detecting moving objects in the image. We have proved that the proposed method is robust to background movement what is a useful property. Future work will include investigation of other features and clustering techniques and implementation of the algorithm in 3-D. Acknowledgement We would like to thank Professor James Duncan from Yale University for providing the ECG-gated MR cardiac images required for this project. References [1] Haralick R. H, Shapiro L. G., Computer and Robot Vision, Vol. I, Addison(Wesley Publishing Company, 1993 [2] Horn B. K. P, Schunck G. B., "Determining Optical Flow", Artificial Intelligence, Vol.17, pp.185-203, 1981 [3] Haralick R. H, Shapiro L. G., Computer and Robot Vision, Vol. II, Addison(Wesley Publishing Company, 1993 [4] Prince J. L, Mc Veigh E. R, "Motion Estimation From Tagged MR Images Sequence", IEEE Trans. on Medical Imaging, 11 (2), pp. 238-249, 1992 [5] Thompson W. B, Combining motion and contrast for segmentation, IEEE Trans. on Pattern Analysis and Machine Intelligence, Vol.2 (6), pp. 543-549, 1980 IJSUz { jk~픇{njCJEHUmH sH jV< UVmH sH jqCJEHUmH sH jC< UV6CJmH sH jCJEHUmH sH jJU< UVmH sH jCJUmH sH CJCJ\mH sH 5CJmH sH 6CJ]mH sH 5CJmH sH  CJmH sH  CJmH sH mH sH * TIJSTz { Kij$`a$$a$$a$w0jD !j$k$l$$$a$    !"#6789DEXYZ[\]pqrsȻԯԖ}pbUj CJEHUmH sH jq< UVmHnHujCJEHUmH sH jW< UVmH sH j CJEHUmH sH joW< UVmH sH j CJEHUmH sH jcW< UVmH sH j CJEHUmH sH j;W< UVmH sH  CJmH sH jCJUmH sH jCJEHUmH sH jV< UVmH sH !s  biqrsv ,-/GHILWXZ[Ÿ6CJ]mH sH jCJEHUmH sH jEX< UVmH sH jCJUmH sH CJ5CJmH sH 6CJH*mH sH 6CJmH sH  CJmH sH A     6 7 ! ! ! !!!Z![!]!b!t!u!v!z!l$$'''''((̺̺Ԯ̺̺ԙԐwj{<CJUmH sH j)UmH sH mH sH jUmH sH 5CJmH sH jCJEHUmH sH jZ< UVmH sH 6CJH*mH sH CJ]mH sH 6CJmH sH  CJmH sH jCJUmH sH jnCJEHUmH sH jyZ< UVmH sH .$$&'''''(( ((N(v(w(x(}(~(((()))))I*` ^`$a$$a$( ( (x(y({(|()))))I*J*h*S-c--- .B.C.J.K.l.....!/"/)/*///0W0w0 6mH sH mH sH  j-CJmH sH 6CJmH sH CJ]mH sH  \mH sH 5CJmH sH jCJUmH sH jCJUmH sH jUCJUmH sH jQCJUmH sH jFCJUmH sH  CJmH sH %I*J*h*i*P-Q-S-c-d-----k.l...J/K///v0w0$a$$a$/ =!"#$%qDd B  S A? 27I*rrg!Dp`!7I*rrg! d}xRJA\~ V1MP4)Hz%b"B `kmoU qaofٙEp ,\ -6!2!3Βl ZKaB)WWa=]+[Bcr3C_rUuTKTckQXd`!aIpo)]9+e d4Qw"u{wYnNZUvM 26w t]}enUoiBG}|6!wv*(s}40 ׍?g)Ɓ%ATzCᚇ>J)oGQW"͐ χ4,qdaDdt P  S A? "2k`FnPp`!k`FnP: q@C_xڕR=KP6m$ D;`ZP0VhA8 ]GEwo>HMx=r*; X~jUwZ( Rl0[qI΢ڭ!V8P;^o}Uo54{' UT~P)Az:g@Ce{4$n=G:>2oMcVS~S:s$Q|4c1ӾjK]v^$ޅGpS9db_g|;L}$ɢkWs៙U2R|Oa1v:!䏒 F,;cD4&"BDC ؂h-%+J(J<>f.ga;s}g; , @\ >,2&3 6V#3BuFu)tMG%@%pn ^jR00^).xGU!zVޞC:SDY]'\(7r(e[N@}g"8#~~̯C McŧOv i_RRMctJɊW-dX ҡuO\b?Xg4:U7e1-,V= KDdB  S A? 2Q 28q^ϻnb p`!ZQ 28q^ϻn d(xQJA}3{'buSNKSTbqDH@@?/ 3S{"D$+X@lk[٧m[9`5^8 pCݕu&$+ɞ%B{39-&W6 uX=r<3mJ6 7OOHE>N]z6*CýBX0@[|qp)l*)%K^u^Y\n FO?8 GL$_/~ )=:|:+_'>tV!.MϦۋ~kUkEv<9 =Q+0V~)?c91w4O3jr  !"#$&'()*+,K/23465789:;=<>@?ABCEDFGHILMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root EntryA F@]1@ 0DataFCT`H GCCC%CECEC%DCWordDocumentEC@FCF\HHObjectPoolCCNYH8XC>@]9_1015698762F ߊ ߊOle CompObjfObjInfo "%(+.147:=@CFGHKNOPRSTUVWXZ[\]^_`abcdefghijl FMicrosoft Equation 3.0 DS Equation Equation.39qܘkIvI "E"xdxdt+"E"ydydt+dEdt=0Equation Native _1015842883 FOle CompObj f FMicrosoft Equation 3.0 DS Equation Equation.39q؀PyII E=( 2 E c2 +E b2 )dxdy  +"+" FMicrosoft Equation 3.0 DS EqObjInfo Equation Native  _1015699073FOle CompObjfObjInfoEquation Native _1015699161FϤuation Equation.39qdIoI E b =E x u+E y v+E t FMicrosoft Equation 3.0 DS Equation Equation.39qOle CompObjfObjInfoEquation Native tII E c2 =("u"x) 2 +("u"y) 2 +("v"x) 2 +("v"y) 2 FMicrosoft Equation 3.0 DS Equation Equation.39q_10156992591FϤϤOle CompObjfObjInfo 8,I~I E x ="E"x FMicrosoft Equation 3.0 DS Equation Equation.39q8ܜII E y =Equation Native !T_1015699299FppOle #CompObj $fObjInfo!&Equation Native 'T_1015699311'$FpOle )"E"y FMicrosoft Equation 3.0 DS Equation Equation.39q8II E t =dEdtCompObj#%*fObjInfo&,Equation Native -T_1015699346)F9Ole /CompObj(*0fObjInfo+2Equation Native 3H FMicrosoft Equation 3.0 DS Equation Equation.39q, J\I u=dxdt FMicrosoft Equation 3.0 DS Equation Equation.39q_10157822576 .F9ĮOle 5CompObj-/6fObjInfo08f,p~IzI v=dydt FMicrosoft Equation 3.0 DS Equation Equation.39q\kIvI E of = Equation Native 9H_1015699525",3FĮήOle ;CompObj24<fObjInfo5>Equation Native ?x_1015700089;8FήծOle A(u 2 +v 2 )  FMicrosoft Equation 3.0 DS Equation Equation.39qx~IuI x i R k !d(x i ,u k (n))=min{d(x i ,CompObj79BfObjInfo:DEquation Native E_1015700217=Fծޮu j (n))} FMicrosoft Equation 3.0 DS Equation Equation.39qIoI u k (n+1):d(x i ,u k (n+1))=min{d(x Ole ICompObj<>JfObjInfo?LEquation Native Mi ,y)} xR k  " Oh+'0 (4 P \ h t*Spatio-temporal image segmentation using rdpatSven Loncaricl ven Normal.dotiSven Loncaricl 23nMicrosoft Word 9.0g@^s@ @Jr@4䬅IHDdlB  S A? 2o]YN+Ldya(KHp`!C]YN+Ldya(xMQAJA]Ŭ b]4y@".(FȊ-xKXKzv&q.O! ]׵?DW/x*?8\|VE4$wvĔ'ptG_*&:%yH8"h@ʺŚZ:t*1iVylGMu6sRY"ckiӨSb;ah j ׽;3Yy]C *ي J_?1p3nߐThڟI@˓+?~< #.t^x r{GŨt9G6I9zս*e;n}aK|? DdddHHV   c 2A 02.jpg R 1)m pF 1)mJFIFHHC  !"$"$ dd }!1AQa"q2#BR$3br %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz?فupG|u:AO#0$(=I=뫷m9a7n/fV6MM>gƲc'U|ן~KNv>V2Wgdϝe Ff8954C-S\BP bU<ªk#nC(y{VOЎrT* &coDQª>&Xl> @82?ۈծ&f('W#$h62 >^swrM#S]֥eo_ myzRnhJ;7φKo ~lVV/e_SƹFg'M͍ѨUO@No,Im.2d0 ˦i"%Bս]}[MNelcl̇8ҸB`h*}cn}*;6r1SeG57ڦW]}ڍy߇aս;_̒yp[Kd]NsuEwo.>~XTYb|i4bd@?TSK t`cz $IR2i@2aYIy=j8xcP% f@7& kd#͑NlOgZբHq`vEc;Ii]&Uhfc!g96=w$~q+!/Rϭx[ʺ2W1cy֤)]G=l;QRRF=A2mc#HsXc_6Xter1?k-Pa9&%UOv_ @H<]6p`cffljORAʎj({Y^۩˖O\a@e8?SPbz[kAX\[W1Ɛop J۱>|j@jF#ne8`[zRGp.ƴ@De>l{VZue'S\Md5y,.tҮNC<8~ z e/[m큟QMN,b{ܝn>z H[cs%*PX͢M<[Aomxwy ;ytw&[s𕈰 W dڤGh2(\Сv:\:ԏ,l}CSI¬!fj ^~%E6kmSl5- FϑnRs+{π0M-`Sf$Iͩ9K){v1T:@ld!7>H h~{? ^ŴRk81 HomyPjS3>H]]Y)/+\(.>N_DBtXyEXq$6Q3 F+y;}zZJ<ߠ3~01r >|(K=@$^@j ~ЗdN^ LfV(o{\AJ"x-7_2M"Ƹ_#"bǺ_b$/u}c~tPFyJ$hLoIO R}3m|~e♃Qjee?bj*#\]Y 2wc^f ̅07R.aHSyr7H}]B]wU޼saW4  pm Z?5xFl9 )3iIp;;('&>oݻ"xG`ҁĀ0 R'`yG ڕz@ t$THTi[J! h=|Zutjq޽ONW;<FbV(Y]~թj{?XVJjLKz{a4Uׁ![_~0 )@X>xׂo$"#b„Ջ sXTN2,.d ܔ?XԘ܊WW0Y.{mZKzzyfc5-G/- ߘ 3ij>_/= |;PW)eUO^@8YMW$jN_$ƶfp+8}۟|Bvhd 't,t8Ϳ#5@TdDSPa z R ZY~dK.{XmXŅgf$GlPWz0&* a{sl%N'FÜX_IV\ܿl!Wpy2 .`xRPme>GhqeLysG"ei]V'HOobjX_9ﯩ˞L&M -/gaq`.?mG]ŝ m&ȞZSi#6hNeqܿi>dR#ސ'+y #D"sYvsV­=эf@h/V IjH)˝LNm.;U/gjJxzo Y\>8hV@]8׽OR])jHBYĀ,'+H.,PSIbszֿo\;sם6{cEv !MMxī],>?鵿Q[)Ty7[J :=I{J2&L ÃY7a,JATl^?iMO__ߌ289D>^ +r[}f*5o~uαS4K,~;,ì!]kj$ŬߘQ(OXPSkǸ<^̼-wl1>Ji^bE(֐l'-ûZv֬o*aD0JzVf\=Q6Ig/~<9PXx0 S\^mZhdw:*SWm;^dn#ug+uZwel;d>"١.yMu %%m a^ th8տO.VG2%Pl\;ciK.]-ċj_Dm/uw f&v|J|>//5]N^%FiH٫^X}!pZ-ł, leͭmꚂf+ 뷂{7ƧI^XvaGS~ ]'9d _md+~[bNKA@4Y&JG*=W=ACM"hE! =hExҬQ}M/W(>MlςHɚ I+.5}* *0._Wg$,OuXDɖdL10{k c1H]/ob.;'ҷ2ܭp@jgLʂ)d|cDJV­iVҔʫ Eorð7<' ,VŔUŭDgb5jjD n%DoD]}t$ċrIA<aE!|$mTWJ~sŭ񲕰Q9yAZOx8) d42$_mi ZbTQKAGu H[Iʸ,TsAR)vFVT[GwVD> ұ p'9˾sNv2Hī[e]MWAryBeks+G2U@ʔ1ixT|IAJSɉg3R%0x 8)Ép+Ցt} 'Z%jAaȀͰaIENDB`Y Dd*8%))v  c RA.cluster-bez-pomaka.bmpb /.?҆vm?.k <pnc /.?҆vm?.PNG  IHDR*Qc!xgAMA0PLTE^`IDATxar0ahN 8ǖ,KO۪~yq}= c R R`7 v R [+|Ȧ[+\C R R(m)m)m)m)m)m)m)md|>>D" m)mH6/)m*6Wņk\LAJA|AJAJf~ͶW؈6Hi6HiL6d6Hi6Hbc,6kh6Hi6HdL6i6Him6f5nHAJA|AJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJAJx<ϟQ(6XCAJAJAJAJxaz;؋AJAJ R Rmݦ.g6\ m)m6X#hj5N)mf?/qj=/]#!m)mf7n\x4{ i6HiԻJ\߂oxkpi6Hi6H5yQ\]ﶚH~5A*F8C R߅YyvHn߅y+&_#)m fb/kh6H!m|'/ ic#ec/~k/I56. _CFsW#X6Hi6Hih6b3kh6HiT$7hS;oW׈dc5AJAJx_zkh6Hics,Qqvm̻6.OKkl&-Y#]zN4s#]_6w}(:khH{ˏg۽kTgA6OYӻDX9iژt mrgm>;Ou6ͳ6N[󬡍Ӵ6jf5jչJثdG_ ̎ߍ Nh m$ͳ6YjV.RY}gjVk6NWp:J?SB7/>}E6\#5qxgK>c m\>^khRX|cU"G+ZF=X56HiT7gds>鳟{Ķ6\$m)m)s^K~s R R4C4\CAJA :U_.6\C R [uxK,u;i6Hi6Hu:UUn?cӺ~ 56i6HiT0K,uޟk)m)mnVuaAJAl#Rxlc5AJAJ6R}qZvd)6Hi6HiTt nA i Rژ4m)m)mF̯]j# R 5j#f R R R R e(66 R R6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi/>~^jIENDB`( Dd(D%04n  c JA&seg-bez-pomaka.bmpbf $8IPeG(B Gpn: $8IPeG(PNG  IHDR(iMgAMA0PLTE^` IDATxan6!G߭[ZWH4$g*>hj'?_ئ6-RB?>m~ߦKiS>pœo/uPju@8uf.]oٳupuPG˿w)uLGAh+t8Rꨚ:P:P5 7wܞ9qZ@ThƒKɟ"Ls䤎ԁJԁgq:IN%#?uRut4EPpԁJԁJ ;;GsLԁJpR*uR*u:MlF~4nduR*u姟y%l lduR*uR*uR*uR*uy/Y!r:P:P1UcVGsTJԁJpS*uR*uگ}~qB@T@OQ}:ԁJԁJ}!FLԁJԁ35O4tuѭzS*uR:gyS*uR*uoiQ>FR*uRF{p@@T@TwW::P:Pt8T@T@嫁|:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:Ptܦi(6ԁJԁczwԁJԁJ^:ڟX/P*uR^^ O$@T@TXI@T@u/r r瘧T@T@u LR0G#\9>T@T:rf8q& [:P:PL$a:#'uR*uQxuP*uR*u?8s:PU_U}xspԁJԁJбV_4cP*uR*uB3Js:PFc3FsԁJԁJ:~߁OHΡT@T9$q(IԁJԁjT)RΡT@T@T@T@T8)r?MstƝCԁJg*sD~Uj9ԁJԁJ0調6K72E ΑR*uRPc9ŧ68MsT@T@c-qəd9usԁJԁ canxxT@T@q )RHΡT@THSTγqK]#Ա].pul8: r|wuS_.19Ա^%ձ TpIu#\R;nHJG.=FR:P:PUYNkTY9Z@T@un=wTS.v:P:Pe?gg첟 tԁJԁ+7?WԁJԁJ.VgH]sԁJԁ cO}MSs:P:P5#4s~wgjZG9QifCefKX9gռPv@TBW~cuR*uRft 'IO*z#PjuPjtx s^-US:9:Ps5:Ps5G:2zg_7wqW~A Ρ@͡=H=~ޯc9Ա:P:PC\:8: CΡPG~'ORG~s#?uR*uRh:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P:P}-xiIENDB`Dd)%66t   c PA ,of-seg-bez-pomaka.bmpb?9 oL'@Rpn?9 oL'PNG  IHDR)WgAMAPLTEٟ|IDATxKn0a,9Gjs4#tM"a$߿hJ'$yI?YQ^ C$I$I$I$)jɨC({02"23$Id0  Iҩ,? #`0ﵕ0R3"3O#q-b+ᬊ03я1r6b01Fb0V}͑yجLdQE`/]Z`3&v:v124n5{3#QE1~ܨw|W#1 01>sFY03'^FyI01#1;Xd|bpio܏}d0ŀh8-|t#EX ECVadkagmJojBݹel#F0Smz7[_f3j4}z5b.&YS{sԽMG+;#1A pozicija.bmpb/;%ٶIGP kx ;Vpn;%ٶIGP kxPNG  IHDR:_5gAMA IDATxə8-a*qi]Hѡ5l}!@wuFsu!@?s pO t`H! t`H! t?r܋?:0C@: a2lD~Џ CKGȠ6>XFj9cB?Q9l6Is@\}A@_e8ȫOd!Xl4 Mp H3$$sHC9c@: 3 y`u_SP:lJ&+ ":/0=y^3 zxH= t/̢!9 I{a4 fPa<(AM:? t/L9@:&qWsWp1w z`( Æ^=p=!M.A[Jzzn)빥tH! t9$-lPMU~/`mn1NC7E #ŪӋOOyk/(,s?9Ū_eɟc|>8v[C0ti\-6qVGfG.mt3>3ے8lp9v[ L ˥<L Z"]XaC&<ϤBl.uغ33nit[V:CT Xlu/t[8ֱoH܂umnn&@: 0C@: 0C@: 0C@: 0C@: 0Cl%+_o}5WWl#+_͌B~qخ|ޯG:4Hw }s;~>ҹxfp+ t`H! t`H!ܠ"&"o"V:Q@A(taH0A0҉_7N@2t;~,IJ B} t/:GvI఩4l"QF6RVGP: gtSVI DVZ0mY&X侬*9$9-B(KvuF*ObM5A{Np$>%V'7ڴGrg_rf [s=չOdRvgst0Nn#MٳԌ 䦾*8žҡS'!tǞ9)}B1 mȞS: ܾ]S:xJ֫'5е# ~\eе!NN4Hžixml3ku#>My\SZ4&܇_?ss1KeP-ǚˎvtpNHT_`1 -Kބc;յW :)Y;irM=|R<6ΝC38!+pY$yZԂ<I<[ V'bi=n;i@t qWOwҡҵC!!!!}~*cg?Ǚgi56$^mMp6~)|a^(;"$;P~( cu:RftRIv,%Á kt`rFxtNE X}ytNo^6N*A xNrxK]ZtQA\*f8.p8Ÿ>x!:ZC>ӸL:Lo}hH;JZIk9%ԗIgAhbvwҽ+Iyb ij82l_<ݔi@] K Z׬uCڸ$թ3<ץ DͳVwt,_C s.).̰i]S1̹|N:s[>6Dm7uȔğwgu6l۠;\q]>A:A5NO]mb]PBl%ܚs;Q?]A|7XAB(͔} WxZue[J YMC)ȕ=Owj=:U3;hx>e<\}8AqΏVΖ̵:z꬝#8L8g[ku$[ǖY$}s2IZޣNe_:&er` Vd؊-9w3lMHƕ^ACa쟕iNby>ًiņ\QdA9 ?g%oHp­ R \3;ֱ,nA^X-Ҷ|0XN.&nm}O^pb7Ʋ$NQJ N=x#stG?3ff[}|_`p`SsSfZj`gynwT(z'GTo ZYTW=3{O֧zS[<lZ} _"֕RNG7[ sSj6^H|s}4 N 9OJޒiȇe2|,sz26/Gxt5R~ + e=(*"+8dCS4 ҙ&$<ωme]3O_'c4_rUk_OW:[~B`z2J-uNNӴ)m9K{t]kw>lZgQSX@I?gi N%uQc}^ɲOYZ&Nn}nwg2\@O<" fg^(ؽ6iQ[tj91+4EMJ.АzKWmK9ofb;<) Qp fsMk#Crz3l]<ۺwv1RiD"j> Yߴ_m_L{.}.Ө$rX g|y޽°:/tg#?ّ> Y'YR-ś5>*o¨;PU:Y& *R߮ uI ؊,$ ܿsPDO 6UD/ŝ_PCXruѧ&xVH,έ%_e ݂26{:=W2upNg }tw>PMДޙQ }AF ¦z6#w>0fkױn%LUgH&FYΌ7&_'[T)=DAF"X9k甆NZ-CS3jF2kg RL/s x=蹔o)(RΒTW\(Ħ=jl cs]R2}ۮ;֣JHޗZ$TG53z.ذt"]y3!M}X00 n5۟q2! xgScrmOw]8:دy{,uc7 %zjkΰæEѴuvڗ }8KUDrOE_'r~ŏ]ul"#+'Ɗ?hx-"i#4OcrqKlc{|Nt"۶>( =JMfY¾K;G'+T rcgVi^qbaP=?) qotfan;ىO݁IIܵG^ [x, JV!Yz?F'y+Ƭ3;3C: >+t\X'j_tL^#}+a`W|r*wfÖpGɛ,6H,DŎ<U󳫸ⳭNkB ]s9:PpSbt'0W楡B +^؇IV19sGmSY]QE^A/ͧ,-zp|CddL-ơcn%ZfaYJVyZȨ+4D8e^iDž eH'o!*zۣYY%Dž %y4ћ Ur of^חéH4fy¦腛%"7E˅N']dP'UV!7iT~iklz`%@:޲Jz^;;BI"yʝ{uCg)=}Ms}[cH/SM:IHK#uB}Ƞn.luVdcR{;BQN(o"&I脙m}ͻ[t&z-$bG;cB}*/=k%IyD3BCDcH;64"TO'켠*B85]=~vnmr`hiLeK\!=pA"Ufd:]”(?f_Z1SssoKX_)d!"/>LLXU^%'yV;~SR;GxoJ"IGȠIyO.^'K^hz[gDԛQsؒ$)+}*"MdtC /"@6XqKG-gکڜyRـ1mɎ D9~b,B-e;ڬcه×Ndp^t4wh6P0u8~iGM-;ٖ#T׳3)ƻ׻蜗Z4md^E&c&׎vJc 73N4=m宝BmoH0_M#ybio`rBM nuG rWAfdgθ#covS{fcuL~i|rK˖7R'^ Z0#$]ٵzf>-:כSZcvxv%Zu<L3q­0N@}fvD:@1M~/0=4AM2X#_Jm61RaS7fq\gbz$Z&%S.UڨR" ,ޜ6swHL>>0QlN!7fѸ dKP|۸'C+;;\RMTE# L)V.E9ru:͎߾ŬЬ%,P ;Olҭ3Mi!v+I&+~;n}` 6Ԕ}!w78p>v;mFzl׏z륟Pgiحa qW1Zev[&! C_m.loH3BBd[Pz2loe 5@x|.quo fvN"@CjKV2SFGPgis1[ ឨA;!* n}SZ MK- lzx_v2i ,jXda{ 7uoss?g`V,: Ғ0rѽ'!M,0#"o^AUZѥn 2Bq_M:W:hc  ^m^"BxeEyq[ 7K}E̹yWRKCiB:!>ےY\|/psmInmtqqtyL '99~ir†;. 5#͵rv .^/Tzt]{|nטdt ^1Ю')*iH5amӘztV񯿎=_)MEk T6.r5& զɔ֌޲tMrdk֜BbKl㖨PFKEZK=l=:Dx1.фpIgq* AuZ/""ph>&7y$pRa^&8m\+KLr@C49/MLa=:~(TgM7QmTc <')ffO66F77A6M*RsFߒ{?Xu(դ=y^c;.qq%"2o!uj,QCkKOI TWqii$›;`<S f\##MIJu*c+ů.ljDxTSԬX-.4wX0$6DR8׶>, b:9jfUD_ܔ ?~Q{+$Goڛy .1 _Cn aboa$ LE)Eꖈ@Ğrmq ݯT?ϣ5:_9m&T `uT rAэ[aO3&]׏{uIkxqDG;~nzBJnqVY3hjt&=ٸfܭPNu_bg"yE5LM^oŨɫO)$pm.ƻ57;Dz6gW%JdЦӹeU,1Pz,1z[2v˺(IpD5{7Wl=nRilt^ DΨƞ][aVv2RM؇3OHDK莧P :nYk릜Ӂv^٦Ga}*qF ;MbRqS+#V`kLb$6fb|½\XTLYc~STd1]Dqr}&iEN d"k^3[)Ö?̈́;dƏ+c&GI7S ;oTҁQ+])Xg%eyZuN Lz2MV@xD\WNAn80Ta#Jk$Y,{Au՞;ZgPu/ji*ݴ12.4͆/)RY5bf?=ԯ4iͳMNرGMuj%9; [zt5ޮұO6C~[q^U&ۉۯsԀct D>#q8 چ1@@ĶXbF.y=QÚͬNrt۪z -^Ktfzm&< eßѹSar{)1ۉL%FNV~vae T4bp .nqªdg5dt'f.ɨIT;1=j@ÖQN>bKҵ^hxf;/;D9SqF-=}FnT (<^W3 Ŏvb. D*LpJ%ԍbРr-q 3=XEs_b4l)_sF7f@fYh'җaKҬl%*  lT),Đ8/.6H8DgSNfmf&EtR-&-f @&B~&щd?S2ϝ.neɂnfĀdՙ҉E m[͌~1\)su:q~?x/ M?7e%iǫF<֦"ߜ\v bڀ~tp+)tHc QV(5*ER*h+)*?dΰ ر6 ql{NFJG.u Y"kClw$wߎX [ $"̙OfR|g`[I ʡ),띸,C0%&1iVU]&fRJH)2 ZS+"K/6ö6 jf4 rNrz1Ԯd9a!׋'?S[PTӚvTRBxIQk4>q)([kƮfw>Rgh=~ %cD1HeCp}>ky&Q DzKk߯ze<}5'PX}ɻ0 |+{{sLg:ᔠzbj;s -/+ Qf ~C0U3&&(Dsl"ZìCK{L 'R1ڳ"1[B|FI$7 !3GZe#E I~me'ҤŅ={i:6s3:d&qt/EDCҊAP䋅ڢd]dklO >(B"hL~3j3r 1bA of-pomak.bmpb8x!v8.1q8pni8x!v8.1PNG  IHDR) ggAMAPLTEٟ IDATx}ߋ+ٝ $o`/?bcOwc-zU5yy؀ k=,1֕< S rI[ժϽ.KRUS*N{nR}s?Ge@P&$) :@?9LdZ7XG~DfBfE(IȞ_9vckL sRc i"o rQU FWO P[lHߦ* jR3%F-| /16`Õ^/ee9 d0^(/#<՟w:xy_s@(J ;``:xaWvx4*WP^E3 YրV!h!gّ-B'RtFU$0B桺OfuQկ~4JuYVY:KS ,dF ƝqQA|J+FX7+Cl.|7.>hV$ }1('8s\]{ܣBڊc-G5E铼PHsJv*뫢8yrRE+yä:$^}EY_LƚW2+FYdѱ O# GwY"9T Y$u>q{۷!y$(Wʬ]'_2."xt<ﹼh~f"j1o@9u,ϙuAMNe] Y_k[ldǚJW5+b MQAOB=.2H$@q= 6Itrgs3\6>άTZfH͊J=%iՒ s>'<>zfyā {ݣۭ QЄ oyPシHeftA@N c޳ظ3HÈ\53@0cdĤ`$ V“paەu=B#+d4Kp -RR얥!'aK7p\AX/5*W~PAvxKy;)%R}b:p˄D#yi fRWp յiT֫WxKA=ts' UWŽUȺr0$G0=4LE\TJ0!hڂtٲ_Ve{?t `sOV^ cϙZ3/ _RH5re^^nVYqF=CR ]wӖo/D|IG+)!- $!r`xv껟*^Ѭ.srQ]ҨW{Tgë:P1vU^h!4u^ ew?LY 5;{H {u$ᚿ;T&^V]zW0 `^h#OH^de- :ŬJ1{]?u W.ZvTH^rGXNviz#A|YVwe {#,)]P]; DrcoGfuېb,;cx6cWnzX/SW|G^(1=xC9.fH H/f,T2h[4'zԾ7# Q;ĀKsW?q[vr4N֛RV4NRԊ{DF $x\jKV,y=ˮmVL7%\Cտyɖ4 V(h=]lh} ;YpRQDGն9i@OYE* 28ڬ|Jڈ/֊,1mn|Q PyqXWĬ\-lT#,&^/jc05C푴n9dV\܄+lX; hʬ wVִ`Nsr;r-LnjZ$Lh |q=. I-OnTm6~]O16JȖ!+zb Uޮi€ܝxh݉x|< p#1T( $/SuiNugЂ B3{-,Aߴ'w诩 7N vI r77Y`꺁7/3 \R\X[ǫhMm؀4@(RJA[!m &= @6+J`bNM6}+ v>]tgynyMIʬj>'€r#c|lŅVVi;e )M;JubC* Dv-颇٪ucjVQFX6>պoLк@pfuFdr鬨ST((Jk7` R[njق @]UA2u@<+ӂW}HV]16U3P}ڶ)gBr Z? @dzrAN;Kj@@]+feTCw#5Y*upX^K[FajvC ͊lכ3gi28I޺!j$?U]Aenw9ƪGvVR +., V&>RJ12Vʅ5V$a 6,"hV07En'*+5/2'aZ$w^/W<uM&(F{U6ZQf>;ċ s]Ym0ǗFMOMiq@^d*MeIÓLLj-W8ɔ{*1-L/kηj`#eL/Y\f%k}\M~~>@{*5 ֔9N[_.p79&[hC nmnJt+A[iz)4.Lګ6- b<;&BN>I[ Ф(=n/^ìL<7.*ycm@pEA҅rv@`}#lw0 pFðQT T٨OYM,HnO.$]  >m !cMr*A9mbœŅ:gK!E(J"8@h&į2|)M{N,,K|Y^UD{Ȗ+JQK2z nmMQ*? =}NB˘&77DokQ󼕲Nޮ<ज़E΢#aclFR>5p@Q/GV]OS_9RaHת 0G0ijyfEPMi68Req$ޡ,_/U57OR =~-hVjy$HxUr8B]V.[4t%+W7eU%JLFv. r:#нˤi.u4'M ʫv.PWhKf1KW\ퟋF^aUk%3|J4W\t"sz@woƐ(,wksdOFM^_[9cA'q`$^;ף 1%dY8#?ߘm[M'eև/u1+m A_*uٲ3oZ|O?)D2fQdž 8M<=&4KJ!\? dOU ƨUN-^]O:L=6&S3얋F6[ydGerQ"mNvԅuh8 8)K;_3+X|cVKڂ._0GhUhrsm_w-Ke{Pqb4L&Gt\ 1#&F1H%ni'Nӈg"RXčp2]'wEׁs&k= < F)u:uj/b 7Mz~4/xfXIX].aSU LMWXVf  r)='kU֜\r!42/h8g ]zD@DtdV\u`,DVd`Up ahyw7hEUT'i5w&/tBU'uu PR;1lf3^ .b ?8nPW^tStD/0O5/x-}qx hHpmp@=z_L-dwvIOO]fY< {_ @;RL?73 !Wż;%gA<]/'Nd Dwb:$pX. Ro[C$A$h\@qNPe-ߒIp Yp ^_\ &(njd03O*GwaKWa3cC1"g_ B K</0̘է8rKg΄1P|.΁&c4~MfmrW&ϸWekA ^ua;3Fu6['x,T2RF>_/4GJ\Bs/mؕ(o|wf"uRZfZvM4~eħ8z]"A6XȠvBOz3[ gyH7шr_~hG](чMiY-Z-O.di8m0+=ˀ@=>qC+|:|.pQO&绽pxpLE;Ëe%[ynij(0$.f/IvGJt ֱ 'rs"l,]$ngTt2՝vB?5]''=\zǗ/> p8K3L*3r]/4HUDo?daA4 X918؞Ҋ :X$AՅ ћPg1.;r PfdJiaDS=Lkb -h 7'k)x rA=w hɽ@/u{GyEkBTrPPÎʳg&ȉu.xU JS"@('xص Rn,"uA鎸ËntBOߠKlGJ^U$l80L;DwwdK W_@u`T&~C򂲣E"ɴ Y'W 1sgМ̀`]3 Ga`y22  `n O(9 n9s'V=@8?ٝtǬcK j)xGo)0$S;MVkvŁ}g8UD "/yy/ӓeSGL< 4wq!T Wă. xٖ' kWeW0? 5ynշI2f ]xi ^:ʷ#K<+6ښ$)~dK* 1Um? 0QF0nh fzV\%^ [NgTE+@dɪxKUE @iU'5O}qJ29"7'8tA;V4p딖wbQ죣o'2DslAȘa%$t>H\{Rz:'8dCPw{r*|0z7='`]@B{ 2$ v=l V}7n6dAX RW . R\d6+UC) Yu*"r+[vi2̻Ʉbe Bqx+ie`% 7O7d;0dʵw KJOL=⇑W ٧~+{9̵6&u[ by;N fC<,on "/wq)H{HP}/3x청̫f;7jdHn ^(Bd08Ϫ4s@ܬA~&4f3玊-pZqNכ M$EKg2QJa06Za =2epZaINg0՜Yw9ŧ>/Uo@Bxul);.\ &Ń:]q`YkKYD.\^8 Tq-NpJOׯ@f': __փ9s6$R絹.<%hW@B/S+[7/~ peҕ)tB"I,JjqkIv͓c'/_z;3R1yZ"|.I䓏?lF[\'?=MoooOM1:M QY:g'曗/•yM\{=;¬iQ 6X;P[bQBȻ HR8Pp_jV$kGfnEU "&a(A4/|g w|Bn% eNV%FQ2o ܍SykʲŸª+I87LYAqZxt *6L"ܙN*oWZ $DhJ(`niTVX]8{s zgڸ \琕kNXh/X`ݬhlXğr9fͤ'N|0d_x+U?ҼQ; &Ns CۈGq}pǔI0N$}볎љܠ*!B |G֝V4e8Oh )2S B8بb=/A6oaS Aw:ӢZ&T!hz ,l6]Ҙ6 \w@BtbA'BfGPSx^5u'>ɋz4Ws2NoҁRDx9Z/?S2$~MAٶخVp6HShBϻ AAσױn2::p_hh/-acV6x,M^M@|>$c&ћQzJ^m!i]L&8ÒIRϣrvPG<\F 5+&`]@p4tJ #'q"!0TydLmqN+H7oơQr({MufyuUcLR'A .cs$ Hd-}&xok}pD*]P"%rVX\[\._YV|흛5QBRf_/nȵlk L<8Sĕ樫#YƁ\e γ:I+ _o׻]۬k8&-g-I|$ݹ&{p/PbR %xK<@e'!Z*{o{ɸgD ځO>XI -;*ٵ6S$X1޽"XWK!~ Dǽ+^_>7u5Ã:qB]ZX{AI@npb^& 7 RN}9H Pd ^L6|u_cIGm]E3u;!{U1A O.J+d2 dBb‘i5 ~p4B ',\\*NlcZf,m= BY]2_ZgMuDAu0+Z_m}# u1#j8SJWTWbŬ` #2I)hJz5ima|W)BVF6Ev#~bBbqZC6ꘓ$TX"PօҾ3"U-T$MZk$J~WyފِئuK)?Dޅex/ xьA e&Ly%S~H\zjG )H90[?WEgv-|ATN.c Q12= ]Ս ڕ ˚nc[ *̉56y@(>z¥䡏*2ý ߲y+H.. ~JE'.#:8.):!#T\ϖ0aO`Yb;-g.L@mDRo%{&r ANLor@Ò^]dQĵO'?L6c6y= bʓe{ٟ Čk;D6:GrOAVMxG *^BۥU$ag- VZm   #Z`V(iu 6%T/&|{quu^\ w\] lI)'=HDCl| gJxr&(3GN/=rC3Z7UJ Γ9an?}6PK޳N Aq\n >WnEo%+ Lpn z> >WnEo%+PNG  IHDR(sgAMA0PLTE^`IDATxan۸FaAv6QOU)Gkp1ixpmnWتU R R R R R R R R R gh[$㯡 R R }6%i6HiT߹њp´AJAJäU')mH;Q"DA)m)mF&` m)mRUX 6Hi6HiD64i6HiT?͝U)m)m:m*vچkTLAJm:A6\A R RE6\QE6\Q R us*ܩp]r R R~GQ' R R R R R R R Rn?_n>\6Hi6H7Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6Hi6HiTz}Qḻ6Hi6Hi6Hi6Hi6H}۸?~4pEm$m)m(7(9)ml<_>kh6Hi6H buk bc5AJAdc'~t@l6Hi6HiT{Y2.h6\c+m)mFkd^*"WpAJAJQskXJJAJ; ȕk)m)mFWK-r IAJqosKURnˆktNAJAWgo8k\~5r)m6jܖqm5i6Hi6H`#AXF! R R 5Yo&xSGJ&T;?S0Ae#uK\ x<0րژt m6-kG)mm݊Y׸چk R R uTnk\n564l6HiP΍-pӬkh6HEs;X#ט|lXicm7t56f]#` nVeBDqkpmx m)mVwB* R H.-)m)m`hxW R RDD56D󮡍hc5QzmF%6B18{m6gTE-,($p m$wSi#]JTHWn os|9m{NU>g\j̓k#Yyrm$q3Od5sɉ6>Gsɉ6>GsɵU%O,yrmdUdɓhr3+m\yw9BX߮`kh#?Os[CYm\yjm-V8bV8bV8b2qispĂ]+?r->WeRn5X8\5qkhp Az.w"ȟǓX~cݜkhʴi m\F궬kw H R R E{zko̽6Hi6Hl,נXw m)m6*wK,[i6HiTO%li6Hi6HXk R Rwv?X R R ۆmi6HiZlc5AJAJ6 <a6ZCYi㻹FV Gѹn R 2n 8Z4c R R Rb_fci5AJA*qhcܴAJAJAJAJAJ-;'jeGF˴AJAJ)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)m)mL-bIENDB`1TableSummaryInformation(BQDocumentSummaryInformation8Y`CompObjkj# ՜.+,D՜.+,` hp  ERICSSON-N.T.aL, *Spatio-temporal image segmentation using Title 8@ _PID_HLINKSA0n?' 02.jpgK'of-bez-pomaka.bmp=b(cluster-bez-pomaka.bmp9} (seg-bez-pomaka.bmpO_x( of-seg-bez-pomaka.bmp@{(  pozicija.bmp)  of-pomak.bmp) cluster-pomak.bmp  FMicrosoft Word Document MSWordDocWord.Document.89q i0@0 Normal_HmH sH tH D`D Heading 1$$@&a$5CJmHsHu<A@< Default Paragraph Font<B@< Body Text$a$CJmHsHuBP@B Body Text 2$a$5CJ mHsHuTC`T Body Text Indent^`CJmH sH uw+HPQoIJSTz{K i j Djkl!"""""## ##N#v#w#x#}#~####$$$$$I%J%h%i%P(Q(S(c(d(((((k)l)))J*K***v+y+0000000J0J0J0JJJJ0J0J0J0J0JJ0JJ0JJJJ0JJ0JJJJ0JJ0J0J0JJ0JJ0JJ0JJJ0J0JJJ0JJJ0JJ000J0JJJ0JJ0J0J0J0JJJ0J0J0J0JJ00J000J0J0JJ00000(0((0((0((0(0((0((0(0s(w0 "j$I*w0!#w0j ~   "68DXZ\pr   w+::::::::::::/XR$S~nܱn95D# pb$/]T-|BSp@0(  B S  ?w+\s!!""""""""$$$$O(R()))))_*d*y+\s!!""""""""$$$$O(R())v+y+\sbi5!6!!!!!!!!!""g"j"q"r"s"v"""""""""""$$$$z%%%%}&&&&&&&&&'''O(R())v+y+ Sven Loncaric!E:\paper\ispa00_flowseg\paper.doc Sven Loncaric!E:\paper\ispa00_flowseg\paper.doc Sven Loncaric!E:\paper\ispa00_flowseg\paper.doc Sven LoncaricIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of paper.asd Sven LoncaricIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of paper.asd Sven LoncaricIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of paper.asd Sven LoncaricIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of paper.asd Sven Loncaric!E:\paper\ispa00_flowseg\paper.doc Sven LoncaricIC:\WINDOWS\Application Data\Microsoft\Word\AutoRecovery save of paper.asd Sven Loncaric!E:\paper\ispa00_flowseg\paper.docy+@HP LaserJet 6P\\Printserv\hp 6pHP LaserJet 6PHP LaserJet 6PHP LaserJet 6P@g  XX@MSUDHP LaserJet 6Pd HP LaserJet 6P@g  XX@MSUDHP LaserJet 6Pd P(P(dqtP(P(w+ @UnknownG:Times New Roman5Symbol3& :ArialcCfCfCfI#L!0d,2)Spatio-temporal image segmentation using Sven Loncaric Sven Loncaric