Snake-Based Segmentation of Teeth from Virtual Dental Casts

Size: px
Start display at page:

Download "Snake-Based Segmentation of Teeth from Virtual Dental Casts"

Transcription

1 Computer-Aded Desgn and Applcatons ISSN: (Prnt) (Onlne) Journal homepage: Snake-Based Segmentaton of Teeth from Vrtual Dental Casts Thomas Kronfeld, Davd Brunner & Gudo Brunnett To cte ths artcle: Thomas Kronfeld, Davd Brunner & Gudo Brunnett (2010) Snake-Based Segmentaton of Teeth from Vrtual Dental Casts, Computer-Aded Desgn and Applcatons, 7:2, To lnk to ths artcle: Publshed onlne: 09 Aug Submt your artcle to ths journal Artcle vews: 219 Vew related artcles Ctng artcles: 21 Vew ctng artcles Full Terms & Condtons of access and use can be found at

2 221 Snake-Based Segmentaton of Teeth from Vrtual Dental Casts Thomas Kronfeld, Davd Brunner and Gudo Brunnett Chemntz Unversty of Technology, {tkro, brunner, ABSTRACT Durng the past years several nnovatve and technologcal developments have been made n oral surgery. Today, dgtzed dental casts are common for smulaton and plannng of orthodontc nterventons. In order to work wth these models, knowng the exact poston of the teeth s of hgh mportance. In ths paper, we present a new method for tooth segmentaton wth mnmal user nteracton. At the begnnng, an ntal estmate for the separatng curve between the teeth and the gum s computed and optmzed by use of an actve contour. The second step calculates the dental arch and the nterstces between the teeth. In order to detect each tooth surface exactly, we fnally poston a snake around the cusp of each tooth. Keywords: actve contours, snakes, tooth segmentaton, mesh segmentaton. DOI: /cadaps INTRODUCTION In orthodontcs surgery, dental plaster casts of the patent s denture are made to smulate and plan nterventons (nlays, brdges, malocclusons). Tradtonally, clncans analyze these plaster casts manually. In the past two decades, computerzed systems became an mportant factor n dental surgery due to the mprovement of three-dmensonal scannng devces (for example [5, 6, 15]). In ths paper, three-dmensonal dental models are represented as trangular meshes. As well as n the tradtonally case, the frst step s the accurate separaton of the teeth from the dgtzed dental model. After the exact poston of each tooth s determned, one s able to measure orthodontc features and smulate orthodontc procedures such as tooth rearrangement. Tooth segmentaton, n general, s a dffcult task, snce teeth occur n dfferent shapes and ther arrangements vary substantally from one ndvdual to another. Furthermore, when dealng wth models wth severe maloccluson, nterstces may be mssng. A lot of automatc methods have already been presented (for a recent survey see [21]). None of these methods have been desgned to specfcally deal wth teeth and usng them drectly for tooth segmentaton wll produce bad results. Algorthms whch were specfcally desgned to deal wth dental models provde better results. Most of them, however, requre tme-consumng and error-prone user nteracton. The method proposed by Kondo et al. [12] s the only hghly automated tooth segmentaton method we know. At frst, the user selects four reference ponts on the surface of the mesh. After that, the algorthm computes two range mages: the plan-vew and the panoramc range mage. The plan-vew range mage s used to determne the dental arch, whch s used as a reference to calculate the panoramc range mage. Both mages are processed separately and composed nto one ndcator functon, whch s used to detect the nterstces

3 222 between teeth. Unfortunately, the method produces poor results, when processng models wth severe malocclusons. The approach of L et al. [13] uses a modfed watershed algorthm for manual segmentaton. At frst, the user selects some ponts on the surface of each tooth. Startng at the selected markers, the method performs a regon growng method untl t reaches vertces wth negatve mean curvature. As stated by them and many other researchers, the boundary between tooth and gum s determned by vertces wth negatve mean curvature values. For example, Zhao et al. [28] descrbe a method based on the calculaton of mean curvature, drectly. Intally, the user specfes a certan threshold as an upper curvature bound. Vertces wth smaller mean curvature are grouped nto connected regons. Then they apply morphologcal operatons to avod multple branchng and extract the skeleton of each regon to obtan thnner boundary lnes. In order to acheve a complete segmentaton, the user has to remove nosy lnes and to connect open boundary lnes manually. A smlar method was suggested by Tan-ran et al. [23]. Snthanayothn and Tharanont [22] propose two dfferent manual segmentaton methods. The frst one s based on the selecton of landmarks that are located at the transton between tooth and gum. These landmarks are then nterpolated usng a cubc splne to complete the boundary. Tooth segmentaton usng cuttng planes s the second descrbed method. Here, the user needs to poston cuttng planes between the teeth, so that each tooth s enclosed by two planes. Besde the methods mentoned above there are some commercal products [1, 3, 4] but the detals of ther algorthms have not been publshed. The workflow s as follows: the dentst has to send the dental plaster cast to the company. After a few days the dentst receves a data fle whch contans the dgtzed and segmented dental cast. Now, one can rearrange each tooth or ft n nlays, but wthout the possblty to change the segmentaton. In ths paper, we follow a completely dfferent approach. Our method for tooth segmentaton s based on actve contour models (also known as snakes). The actve contour model was orgnally developed by Kass et al. [11] to detect salent features n two-dmensonal mages. The am of ther research was to fnd connected boundares based on low-level nformaton such as the gradent magntude of an mage. They postulated that any object boundary could be descrbed mathematcally usng a parameterzed curve. The curve s ntally placed outsde the object and evolves under the nfluence of nternal und external energes untl t algns wth features of nterest wthn an mage. The nternal energy derves drectly from the curve and serves as a smoothness constrant, whle the external energy derves from the mage and forces the curve to lock on salent mage features. The overall evoluton process s obtaned by energy mnmzaton. Mlroy et al. [14] extends the orgnal snake model to three-dmensonal surface for segmentaton purposes n reverse engneerng. In ther approach, the snake evolves drectly on the surface. The user specfes a small closed contour wthn the regon to be segmented, whch then grows under the nfluence of an nternal pressure force untl t reaches vertces wth local curvature maxma. Each vertex of a snake s moved to one of two adjacent postons on the surface of the mesh, perpendcular to the snake curve. The energy model, however, requres a locally contnuous surface n the neghborhood of each vertex. For that purpose, the regon surroundng each vertex s approxmated by a second order surface. A smplfed model for trangle meshes was proposed by Jung et al. [9]. Here, snake elements are located on adjacent vertces. Durng an teraton, each vertex of the snake moves to one of ts neghborng vertces on the mesh, n order to reduce ts energy as much as possble. The evoluton s carred out accordng to the greedy prncple, whch was frst ntroduced for actve contours by Wllams and Shah [25]. The rest of ths paper s organzed as follows. In secton 2 we revew the mathematcal background of the proposed method. At frst we calculate possble tooth boundares (termed: features) and mprove them by suppressng nose (secton 2.1). After that, the cuttng plane between the teeth and the gum s estmated (secton 2.2) and converted nto one or more snakes. The energy functonal as well as the evoluton equaton of these snakes are outlned n secton 2.3. Our newly defned external energy functonal s presented n secton 2.4. After the snake evoluton converged, we calculate the exact poston of nterstces between neghborng teeth by usng the edge tangent flow (secton 2.5). As a last step, each tooth s separated ndvdually. Therefore, we ntalze a snake at the crown of a tooth, whch then grows untl t reaches the transton between tooth and gum. A complete overvew of

4 223 our approach, as well as a dscusson on mplementaton detals s presented n secton 3 (see also fgure 1). Results are demonstrated n secton 4, and conclusons are drawn n secton 5. Fg. 1: Snake-based segmentaton process. 2 MATHEMATICAL BACKGROUND The three-dmensonal dental model s represented usng a trangular mesh (, ), whch conssts of a set of vertces ( v,, v ) 1 n and a set of ndexed trangles ( t,, t ) 1 nt where each vertex v 3 represents a unque poston n and each trangle t s defned by three vertces t {(v,v,v ) v,v,v ;v v j k j k j v k }. The set of edges s gven by {(v,v ) v,v ;0, j n j j } where v v j v a normal vector n( v) s defned wth unt length n ( v ) 1 belong to the same trangle t. At each vertex. A vertex w s called neghbor of a vertex v ff the edge (v,w) exsts. The set of neghbors of a gven vertex v s defned by ( v). 2.1 Feature Regon In order to detect regons of nterest on the mesh by usng snakes, one has to assgn an attrbute to each vertex of the mesh. In general, surface features are defned usng curvature nformaton, snce curvature ntrnscally descrbes the local shape of a surface. Consder a contnuous, orented, and unbounded surface S and a pont v whch les on S along wth ts unt normal vector n( v ). One can construct a plane P( v) that contans v and n( v ). The ntersecton of ths plane wth S results n a curve c on the surface. Ths curve has an assocated normal curvature ( v ) at the vertex v. The curvature value does not however specfy the surface n curvature of S at v, snce the constructed plane s not unque. If we rotate the plane P( v) around the unt normal vector n( v ), we receve a new contour along wth ts own curvature value. Among all these curves, there s one wth mnmal curvature 1 (v) and one wth maxmal curvature 2 (v). These two curvatures are also known as the prncpal curvatures of S at pont v.

5 224 Fg. 2: Color-coded mean curvature values for each vertex v. Postve curvature values are shown n red whle negatve curvature values are colored n blue. In order to deal wth pecewse smooth surfaces such as trangle meshes, we have to estmate dscrete curvature values. We calculate these dscrete curvature values accordng to the method proposed by Rusnkewcz [20]. For each vertex v the prncpal curvature values are gven by 1 (v) and 2 (v). ( v) ( v) 1 2 Based on these values, we compute the mean curvature ( v) for each vertex H 2 v. The values ( v) are then mapped to the range[1,1] (see [24] for mplementaton detals). Fgure 2 shows H the dstrbuton of mean curvature on the mesh. (a) (b) (c) Fg. 3: Results for the vertex set (blue) accordng to the threshold 0.1 (a) and 0.5 (b). The color-coded vertces belongng to the vertex set (red) wth 0.8 are shown n (c). State of the art segmentaton methods mostly defne part boundares accordng to the mnma rule [7]. Ths rule states that boundares between parts could be found along lnes of negatve mnma curvature. Fgure 2 shows an example of the mean curvature dstrbuton and n fgure 3(a) vertces wth H 0.1 are hghlghted. These examples show that vertces wth negatve curvature are located at the transton between tooth and gum, so we could apply the mnma rule as well. Unfortunately, there are many nosy regons due to the dgtzaton and the natural shape of the gum. Therefore, we have to reduce the number of consdered vertces. At frst, a user defned threshold s used as an upper bound. Based on the feature vertex set {v (v) } (0.1) H s generated. Vertces belongng to are grouped nto one or more separate regons. Two vertces u,w belong to the same regon, f they are drectly connected by an edge (u,w) or f there exsts a path ( u = v, v,, v, v = w) between u and w whose vertces belong to the set v (wth 0 k ). 0 1 k1 k

6 225 The choce of a threshold value s complcated. Usng the value 0.5 results n too few vertces n to obtan closed regons at the tooth boundares (Fgure 3(b)). On the other hand, when usng a threshold of 0.1 or above, many undesred regons reman (fgure 3(a)). Our experments showed that 0.3 s a good choce. After that, most nosy regons have been fltered, but there are stll small feature regons left. Thus regons wth less than 0.01 vertces are deleted. Fnally we close small holes between and wthn the remanng regons, by applyng morphologcal operatons [19]. In the followng sectons, we assume that specfes the enhanced vertex set. Snce these feature regons along wth ther curvature values are part of the external snake energy, we defne a new curvature functon for each vertex v : (v) (v) H f v (0.2) 0 otherwse The fnal segmentaton result s obtaned by startng an actve contour at the surface of each ndvdual tooth crown. Vertces belongng to the crown are characterzed by strong postve mean curvature. So we construct a new vertex set: {v : (v) } (0.3) H where s a user defned threshold. Experments showed that a value of 0.8 s suffcent. An example of ths set s shown n Fgure 3(c). Unfortunately, we cannot use these vertces drectly to segment the teeth, because there are many vertces wthn that are located at the surface of the gum and at the base of the dental model. In that way, we wll obtan a severe over segmentaton, whch cannot be reduced wthout expert knowledge. Therefore, we have to thn out the set untl only vertces at tooth crowns reman. (a) (b) (c) Fg. 4: Best fttng plane resultng from the prncpal component analyss of the vertex set. 2.2 Prncpal Component Analyss Prncpal component analyss s a common tool n statstcal analyss. The central dea s to reduce the dmensonalty of a set of samples, whle retanng as much as possble of the varaton present n the set. Ths s acheved by transformng the orgnal set to a new set of varables, whch are called components. One could sort the components n a way that the frst few of them retan most of the varaton present n all of the orgnal varables. Components that are derved n ths way are called prncpal components. The earlest descrpton of ths technque was gven by Pearson [17] and Hotellng [8]. Pearson [17] developed a method for fndng a lne or plane that best ft a set of ponts n p-dmensonal space. The term best ft, however, s mathematcally mprecse. Consder a set of ponts n three-dmensonal space and ther perpendcular dstance to a set of planes. The plane whch mnmzes the sum of dstances s called best fttng plane. Pearson [17] also showed that ths plane contans the centrod of the pont-set. In our case, we are lookng for a frst estmate for the boundary between teeth and gum. Vertces belongng to the vertex set are located n the vcnty of the searched boundary, as shown above.

7 226 Therefore we use prncpal component analyss to calculate the plane that best fts to all vertces n. The calculaton s based on egenvalue decomposton of the covarance matrx, whch s gven by where C 1 (v m)(v m) T (0.4) v m 1 v (0.5) v s the barycentrc coordnates of the vertex set. The square matrx C s symmetrc and real-valued. Thus, the egenvalue decomposton, calculated by the Jacob egenvalue algorthm [18] results n three real-valued egenvalues. The egenvector correspondng to the largest (frst) egenvalue ponts n drecton of the maxmal varance n the data-set. The plane we search for s spanned by the egenvectors, correspondng to the frst and second egenvalue and contans the barycenter m (fgure 4). 2.3 Snake Model The actve contour model was frst proposed by Kass et al. [11] for representng mage contours. Ther basc snake model s an energy mnmzng curve, whch evolves under the nfluence of several forces. The contour s represented as a parametrc curve c( s) ( x( s), y( s)) where s [0,1] and the energy functonal to be mnmzed s wrtten as * E snake E snake (c(s)) ds E nt (c(s)) E mage (c(s)) E con (c(s)) ds. (0.6) E represents the nternal energy of the contour, E refers to the mage forces and E nt mage con external constrant force. The nternal energy s the E nt (s) c (s) 2 (s) c s ss (s) 2 (0.7) 2 conssts of the frst and second-order dervatve of the parametrc curve, where (s) and (s) are user defned weghts. The frst term wll have large values at gaps n the curve, whle the second term ncreases when the curve bends rapdly. The fnal locaton of the snake corresponds to local mnma of the energy functonal. The functonal optmzaton s solved usng calculus of varaton, whch nvolves dervng a varatonal ntegral and solvng the correspondng Euler-Lagrange partal dfferental equaton teratvely. Therefore mage forces and constrant forces need to be dfferentable to guarantee convergence. Also the computed contour coordnates are real numbers, allowng the ponts to fall between the dscrete pxel coordnates of the underlyng mage. In order to overcome these drawbacks, as well as to speed up computaton, Wllams and Shah [25] proposed an evoluton method based upon the greedy prncple. Frst of all, they use a sampled verson of the contour c ( c,, c ) where each sample c (where 0 n 0 n ) s related to a dscrete pxel coordnate. Durng an teraton, the energy functon s computed at neghbors. Therefore, the energy functonal to be mnmzed s wrtten as: n n c and each of ts eght * E snake E snake (c ) E nt (c ) E mage (c ) E con (c ) (0.8) 0 0 and the nternal energy s approxmated usng fnte dfferences. Mlroy et al. [14] extend the snake evoluton to three-dmensonal wreframe models. Ther evoluton method s based on the greedy prncple as well. Durng each teraton each vertex of the snake s moved to one of two nearby postons on the surface, perpendcular to the snake contour. The dstance of these ponts s specfed by the user. Therefore, the regon surroundng each vertex s approxmated by a second-order surface.

8 227 Jung et al. [9] present a smpler evoluton scheme to ensure that the snake does not leave the surface. The snake elements are always located at vertces of the mesh. In each teraton, the snake elements move to one of ther neghborng vertces. Fg. 5: For each snaxel s the unt normal vector and unt tangent vector are approxmated (a). At each teraton step, two adjacent vertces are selected (b). (a) (b) The actve contour model used n ths artcle conssts of an ordered set of vertces ( s,, s ) 1 n wth. Furthermore, successve snake vertces ( s, s ), 0,, n 1 are restrcted to le on 1 adjacent vertces s (s ). We wll call snake vertces snake elements or for short snaxel. In ths 1 paper, we only consder smple and closed actve contours ( s 0 s n whereas s s j for j :, j 0,,n 1 ). On that note all subscrpt arthmetc s modulon. Vertces whch are enclosed by the snake are called nteror vertces. At each snaxel s the unt tangent vector s t( s ) s s 1 1 s 1 1 s approxmated, as shown n fgure 5(a). The snake wll now be orented such that the unt snaxel normal vector b( s) at each snake element s ponts away from the nteror vertces of the contour. The unt snaxel normal s defned as the outer product between the unt tangent vector and the unt surface normal vector: b(s) t(s)n(s) s. (0.10) In each teraton, a snaxel s can only move to a vertex wthn ts neghborhood ( ). We further restrct the possble canddates n the same way as proposed by Mlroy et al [14]. Each snaxel s moved to one of two adjacent vertces s,s (s ), perpendcular to the snake curve, or remans at ts,1,2 current poston s s. Fgure 5(b) shows an example for the choce of canddates. The energy of the,0 snake element s s defned as: where nt ext (0.9) e ( s ) e ( s ) e ( s ) (0.11) e s the nternal energy and e nt ext s the external energy. Note that e,e,e nt ext 3 are column vectors e( s ) ( e ( s ), e ( s ), e ( s )) T. The value of the topmost element of these vectors corresponds to the contour energy at the snake element s. Durng an teraton, each element s moved to the vertex correspondng to the mnmal value of e( s ). The energy functonal to be mnmzed s wrtten as: n * E snake () mn {e 0j 2 j (s )} (0.12) 0 s

9 228 and the evoluton s carred out accordng to greedy prncple [25]. The nternal energy generally conssts of the frst and second order dervatve of the curve, whch are also called tenson and bendng energy, respectvely. Here, we approxmate the dervatves by fnte dfferences. The nternal energy term for each element e ( s ) of the vector e ( s ) s therefore nt, nt defned as follows: e (s ) s s s 2s s (0.13) nt,j,j 1 1,j 1 where and are user defned parameters. Large values of shrnk the snake, whle large values of reduce the curvature of the snake. The external energy e ext j can take varous forms; n general, t s derved from nformaton of the surface curvature. At an explored feature vertex the value of the external snake energy wll be a local mnmum. Here, the vertex set contans the features to be detected and the external energy s derved from the curvature functon, defned n equaton (2.2). Unfortunately, we cannot use ths curvature functon drectly, snce ther value s only nonzero at feature vertces. Ths means, that the external energy would be constant on bg parts of the mesh, and so the evoluton of the snake depends completely on the nternal energy. As one can see n fgure 3, the relatve poston between the cuttng contour obtaned by the prncple component analyss and the transton between teeth and gum, changes completely from denture to denture. In fgure 3(a) the contour les completely above the searched feature, whle n fgure 3(b) the contour les completely below. For ths purpose, the external energy e ( s ) e ( s ) e ( s ) (0.14) ext feat press s composed of two newly defned energy functonals, the feature attracton energy e feat pressure energy e press and the. The feature attracton energy, whch wll be explaned n secton 2.4, expands the scope of feature regons by usng a pre-calculated vector feld, the feature attracton flow. The second energy term, the pressure energy e press, s smlar to the balloon force proposed by Cohen [2]. Hs am was to modfy the external forces n such a way as to obtan more stable results for mage snakes. In order to archve ths goal, he ntroduced an nflaton force by whch the snake behaves lke a balloon. Ths force s wrtten as E ball k n(s) where n( s) s the unt outward normal of the curve wth arc-length s and k determnes the magntude of the force. Ths avods that the curve s trapped by spurous nosy feature ponts and makes the result much more nsenstve to the ntal poston. In our case, the pressure force s used as a tool to accelerate the movement of the snake and to reduce the nfluence of swrlng effects n the feature attracton flow. It also affords the opportunty to control the evoluton drectly, for example when the snake s used as a semautomatc segmentaton tool and the ntalzaton s done by the user. The energy term for each element epress s wrtten as the nner product: e s ( s ) b( s ), s press, j where b( s ) s the outward unt normal of at the snake element s and s,, j, j s s j e press,j, n the vector (0.15) s ts canddate vertex. The weght determnes the strength of ths force, whereas the sgn of specfes the drecton n whch the force acts. Negatve values let the snake grow, whle a postve value shrnks the snake. In general, the magntude of should be small, otherwse the snake may not converge.

10 Feature Attracton Energy Fg. 6: Examples of the derved feature attracton flow. As mentoned above, the relatve poston between the cuttng contour and the transton between teeth and gum could not be determned n the frst place. Therefore, especally for automatc segmentaton, we have to expand the capture range of features to attract the snake from an arbtrary poston. Ths problem was frst addressed by Xu and Prnce [26, 27]. They proposed the gradent vector flow as a new external force for parametrc mage snakes. In order to compute such a feld, one has to defne a scalar feld as an overlay of the mage. After that, the gradent of ths scalar feld s calculated. The fnal vector feld s obtaned by dffuson of the gradent nformaton over the entre mage. One can nterpret ths feld as the drecton to be followed to reach the nearest object boundares. In our case, the computaton s based on the regon ncluded n the set and the above defned curvature functon (v). The gradent of a scalar feld s defned as a vector feld whose components are the partal dervatves of the underlyng scalar feld. Here we deal wth a dscrete and mostly rregular grd, so we have to approxmate the gradent vectors. Therefore, the gradent s determned usng fnte dfferences: (v) 1 ((v) (u)) u v (v) u v. (0.16) u (v) Three general propertes of the gradent vectors are essental for the followng calculatons. Frst of all, the vector (v) at each vertex v ponts n the drecton of the nearest feature vertex. Second, these vectors generally have large magntudes only n the mmedate vcnty of a feature regon. Thrd, n regons wthout features (v) s nearly zero. We defne the feature attracton flow to be the vector feld f(v), where for each v the vector of ths feld ponts towards the nearest feature regon. The feature attracton flow s ntalzed wth the derved curvature gradent f (v) (v). After that, an teratve dffuson process s performed, 0 whch dstrbutes the gradent nformaton over the entre mesh: 1 f t (v) (1) f t 1 (u) (v), v. (0.17) (v) u(v) After t teratons we obtan a smooth vector feld f t, whch has non-zero magntude at each vertex of the mesh. Fgure 6 shows an example of the feature attracton flow. In partcular, at vertces v wth large magntude (v) the second term of equaton (2.14) domnates the sum, whch keeps the vector f ( v ) nearly equal to the ntal gradent vector. The frst term of equaton (2.14) domnates at vertces t v where (v) s nearly zero, yeldng a slowly varyng feld. The parameter s a regularzaton parameter controllng the tradeoff between the frst term and the second term n the sum. Wth small values, one can reduce the nfluence of nose. After each teraton we perform an mmedate step where

11 230 the length of each vector s clamped to the range [0,1]. Ths prevents the magntude of vectors near or wthn feature regons from rsng rapdly. The feature attracton energy s then defned by e fae,j (s ) f(s,j ) f(s ),j f(s,j ), s s,j (0.18) s,j s where f( v) s the feature attracton flow after t teratons. The mnuend ensures that the snake moves n drecton of ncreasng feature attracton flow vectors. Due to the mnmzaton of the subtrahend the snake s movng n drecton of the nearest feature regon and converges there. 2.5 Edge Tangent Flow In dgtzed dental casts, often nterstces between adjacent teeth vansh due to severe maloccluson or low resoluton scanners. In cases such as shown n fgure 7(a-b), automatc segmentaton may fal. Therefore, we have to trace and reconnect the nterstces, pror to the fnal segmentaton. Kang et al. [10] frst solved ths problem for two-dmensonal mages by usng edge tangents. Ther am was the development of an enhanced edge detector to produce hgh-qualty lne drawngs. In general, the task of creatng lne drawngs s dffcult, snce salent edges often appear as pecewse unconnected lnes n natural photographs. Only n an overall vew one can decde whether there s a salent edge or not. In order to maxmze the lne coherence, they construct a smooth vector feld wth whch one s able to follow the shape of lnes. For ths purpose, they defne the edge tangents, whch are perpendcular to the mage gradent. In ths paper, we extend the defnton of the edge tangent flow to three-dmensonal trangle meshes. We defne the edge tangent vector, denoted t ETF ( v) for each vertex v, as the outer product between curvature gradent (v) and surface normal n( v ) : An example s shown n fgure 7(c). t (v) (v)n(v) (0.19) ETF (a) (b) (c) Fg. 7: Unconnected feature regons due to low resoluton scanners (a) and severe maloccluson (b). An example of the edge tangent flow s shown n (c). 3 TOOTH SEGMENTATION An overvew of the proposed segmentaton algorthm s shown n fgure 1. The algorthm starts wth estmatng the mean curvature for each vertex v. Now, the user defnes the threshold to obtan the vertex set, as well as the threshold to obtan the vertex set, accordng to the method descrbed n secton 2.1. The vertces belongng to the set are located at the transton between the teeth and the gum. In order to obtan an ntal estmate of ths regon, we perform the prncpal component analyss on the vertces of. As stated n secton 2.2, the best fttng plane s defned by the vectors belongng to the

12 231 two larger egenvalues. By cuttng the dental cast wth that plane, we obtan we obtan one or more closed and orented contours (fgure 3). Each of them s transformed nto a snake and evolves ndependently. After each snake element s moved, n order to mnmze the energy functonal, t may happen that adjacent snake elements are no longer located on adjacent vertces. In those cases we have to reconnect them. Durng evoluton, a snake may shrnk untl t encloses only a sngle trangle. These snakes wll be dscarded. On the other hand, two snakes can expand and collde wth each other, and then they are merged. At the end, the remanng snake encloses the teeth completely and s called ntal snake. Now, we use the edge tangent flow to detect and reconnect the nterstces. For each snake element, the neghborng vertces are examned. If the edge tangent flow vectors at these vertces are orthogonal to the tangent vector of the current snaxel, then t s most lkely that there s an nterstce. Frst, the correspondng snaxel at the other end of the nterstce s estmated. After that the shortest path n drecton parallel to the edge tangent flow vectors and wthn the nteror of the ntal snake to the correspondng snaxel at the other end of the nterstce s searched. If the feature regon along the searched path s unconnected, then these gaps wll be closed as long as they are smaller then fve percent of the Eucldean dstance between these correspondng snaxel. The remanng task s to detect the surface of each ndvdual tooth exactly. Unfortunately, we cannot ensure that all snake elements of the ntal snake are located at the transton between teeth and gum. As a consequence, some nterstces have not been detected yet. But besdes that, we have reduced the consdered surface of the mesh sgnfcantly. The remanng parts are the surface of the teeth and maybe some small parts of the gum. Therefore, we can be sure that all regons ncluded n the vertex set and at the nteror of the ntal snake, are cusps of teeth. Each of these regons wll be used to ntalze an ndvdual snake. Under the nfluence of the feature attracton flow and the pressure force, each snake expands untl t reaches the tooth boundary. In order to avod expanson, when tooth boundares are mssng, we ntroduce the hard constrant that the snake has to stop when t reaches the ntal snake or detected nterstces. 4 EXPERIMENTAL RESULTS The proposed method was tested on ten dfferent dental casts and performs constantly well. The resultng segmentaton s shown n fgure 10 and the overall runtme for these meshes s lsted n table 1. The nternal energy was weghted by settng 0.6 throughout the tests. The expermental results showed that weghtng the feature attracton energy wth 0.5 s suffcent. Snce the pressure energy accelerates the movement of the snake, we set 0.2 for the ntal snake and 0.2 to weght the snakes started wthn the teeth. Model (number of vertces) Model 1 (142k) Model 2 (208k) Model 3 (155k) Model 4 (209k) Curvature estmaton 1.34 s 1.9 s 1.4 s 1.85 s PCA and snake ntalzaton 1.15 s 1.7 s 1.3 s 1.7 s Feature attracton flow 3.2 s 4.86 s 3.6 s 4.76 s Contour calculaton 0.22 s 0.3 s 0.24 s 0.2 s Outer snake optmzaton 4.4 s 3.3 s 3.5 s 5.2 s Sngle tooth segmentaton 0.09 s 0.1 s 0.14 s 0.09 s Total tme 12 s 14 s 13 s 15 s Tab. 1: Runtme statstc for all models llustrated n Fgure 11. Nevertheless n dental casts, where the boundary between tooth and gum s very smooth, the segmentaton fals. The same error occurred, when neghborng teeth overlap each other due to severe maloccluson. In these cases, the user needs to segment the tooth manually.

13 232 5 CONCLUSION We have presented a hghly automated segmentaton method for separaton of teeth from trangular meshes. The algorthm s based on actve contours, wherefore t s robust even n the presence of heavy nose. Through our experments, we found that our approach s fast and requres mnmal user nteracton. Hence t enables orthodontcs to plan dental treatments much faster. The proposed snake model s not restrcted to operate on ths specal type of three-dmensonal models. Therefore, as future work, we wll nvestgate ther behavor for the segmentaton of several other models, lke dgtzed archeologcal models. (a) (b) (c) (d) (e) (f) Fg. 10: The dental model s cut by the best fttng plane (a). Vertces belongng to the cut trangles are transformed nto the ntal snake, whch s then optmzed (b). The resultng segmentaton for model 1 s shown n (c). The segmentaton results for model 2-4 are shown n (d)-(f). ACKNOWLEDGEMENTS Ths work was partally supported by the Sächssche Aufbaubank and fnanced from budget of the European Unon and the Free State of Saxony. We lke to thank Image Instruments for provdng the dental models. REFERENCES [1] Cadent, OrthoCAD. [2] Cohen, Laurent D.: On actve contour models and balloons, Computer Vson, Graphcs and Image Processng: Image Understandng, 53(2), 1991, [3] Dmennex, Dgtal Lab. [4] Geodgm, Emodel Software. [5] Goshtasby, A.; Nambala, S.; Derjk, W.; Campbell, S.: A system for dgtal reconstructon of gypsum dental casts, IEEE Trans on Medcal Imagng, 16(5), 1997,

14 233 [6] Hrogak, Y.; Sohmura, T.; Satoh, H.; Takahash, J.; Takada, K.: Complete 3-D reconstructon of dental cast shape usng perceptual groupng, IEEE Trans. on Medcal Imagng, 20(10), 2001, [7] Hoffman, D. D.; Sngh, M.: Salence of vsual parts, Cognton, 63(1), 1997, [8] Hotellng, H.: Analyss of a complex of statstcal varables nto prncpal components, Journal of Educatonal Psychology, 24(6), 1933, [9] Jung, M.; Km, H.: Snakng across 3d meshes, Proc. Of the 12th Pacfc Conf. on Comp. Graphcs and Applcatons, 2004, [10] Kang, H.; Lee, S.; Chu, C.: Coherent lne drawng, Proc. ACM Symp. On Nonphotorealstc Anmaton and Renderng, 2007, [11] Kass, M.; Wtkn, A.; Terzopoulos, D.: Snakes: Actve contour models, Int. Journal of Comp. Vson, 1, 1988, [12] Kondo, T.; Ong, S. H.; Foong, K. W. C.: Tooth segmentaton of dental study models usng range mages, IEEE Trans. on Medcal Imagng, 23(3), 2004, [13] L, C.; Wang, G.; Xu, T.; Lu, Y.: Orthodontc Smulaton and Dagnoss: An Enhanced Tool for Dentsts, 27th Annual Int. Conf. of the Engneerng n Medcne and Bology Socety, 2005, [14] Mlroy, M. J.; Bradley, C.; Vckers, G. W.: Segmentaton of a wrap-around model usng an actve contour, Computer-Aded Desgn, 29(4), 1997, [15] Motohash, N.; Kuroda, T.: A 3D computer-aded desgn system appled to dagnoss and treatment plannng n orthodontcs and orthognathc surgery, European Journal of Orthodontcs, 21, 1999, [16] Page, D.; Koschan, A.; Abd, M.: Percepton-based 3D trangle mesh segmentaton usng fast marchng watersheds, Proc. IEEE Conf. on Comp. Vson and Pattern Recognton, 2003, [17] Pearson, K.: On lnes and planes of closest ft to systems of ponts n space, Phlosophcal Magazne, 6(2), 1901, [18] Press, W. H.: Numercal recpes: the art of scentfc computng, Cambrdge Unversty Press, [19] Rossl, C.; Kobbelt, L.; Sedel, H.-P.: Extracton of feature lnes on trangulated surfaces usng morphologcal operators, Proc. of the AAAI Symp. on Smart Graphcs, 2000, [20] Rusnkewcz, S.: Estmatng Curvatures and Ther Dervatves on Trangle Meshes, Symp.on 3D Data Processng, Vsualzaton and Transmsson, [21] Shamr, Arel: A survey on mesh segmentaton technques, Comp. Graphcs Forum, 27(6), 2008, [22] Snthanayothn, C.; Tharanont, W.: Orthodontcs treatment smulaton by teeth segmentaton and setup, 5th Int. Conf. on Electrcal Engneerng/Electroncs, Computer, Telecommuncatons and Informaton Technology, 1, 2008, [23] Tan-ran, Y.; Nng, D.; Guo-dong, H.; Xao-sheng, C.; Ha-hua, C.; Wen-he, L.; Qung, Y.; Pejun, L.: Bo-Informaton Based Segmentaton of 3D Dental Models, The 2nd Int. Conf. on Bonformatcs and Bomedcal Engneerng, 2008, [24] Vergne, R.; Berla, P.; Graner, X.; Schlck, C.: Apparent relef: a shape descrptor for stylzed shadng, Proc. of the 6th Int. Symp. on Non-photorealstc Anmaton and Renderng, 2008, [25] Wllams, D. J.; Shah, M.: A fast algorthm for actve contours and curvature estmaton, Comp. Vs., Graphcs and Image Proc.: Image Understandng, 55(1), 1992, [26] Xu, C.; Prnce, J. L. : Gradent Vector Flow: A New External Force for Snakes, IEEE Proc. Conf. on Comp. Vson and Pattern Recognton, 1997, [27] Xu, C.; Prnce, J. L.: Snakes, shapes, and gradent vector flow, IEEE Transacton on Image Processng, 7(3), 1998, [28] Zhao, M.; Ma, L.; Tan, W.; Ne, D.: Interactve Tooth Segmentaton of Dental Models, 27th Annual Conf. on Engneerng n Medcne and Bology, 2005,

Active Contours/Snakes

Active Contours/Snakes Actve Contours/Snakes Erkut Erdem Acknowledgement: The sldes are adapted from the sldes prepared by K. Grauman of Unversty of Texas at Austn Fttng: Edges vs. boundares Edges useful sgnal to ndcate occludng

More information

Fitting: Deformable contours April 26 th, 2018

Fitting: Deformable contours April 26 th, 2018 4/6/08 Fttng: Deformable contours Aprl 6 th, 08 Yong Jae Lee UC Davs Recap so far: Groupng and Fttng Goal: move from array of pxel values (or flter outputs) to a collecton of regons, objects, and shapes.

More information

CS 534: Computer Vision Model Fitting

CS 534: Computer Vision Model Fitting CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust

More information

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,

More information

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism for Nested Loops with Non-uniform and Flow Dependences Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr

More information

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges

More information

High-Boost Mesh Filtering for 3-D Shape Enhancement

High-Boost Mesh Filtering for 3-D Shape Enhancement Hgh-Boost Mesh Flterng for 3-D Shape Enhancement Hrokazu Yagou Λ Alexander Belyaev y Damng We z Λ y z ; ; Shape Modelng Laboratory, Unversty of Azu, Azu-Wakamatsu 965-8580 Japan y Computer Graphcs Group,

More information

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points; Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features

More information

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng

More information

Feature Reduction and Selection

Feature Reduction and Selection Feature Reducton and Selecton Dr. Shuang LIANG School of Software Engneerng TongJ Unversty Fall, 2012 Today s Topcs Introducton Problems of Dmensonalty Feature Reducton Statstc methods Prncpal Components

More information

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth

More information

Support Vector Machines

Support Vector Machines /9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.

More information

Active Contour Models

Active Contour Models Actve Contour Models By Taen Lee A PROJECT submtted to Oregon State Unversty n partal fulfllment of The requrements for the Degree of Master of Scence n Computer Scence Presented September 9 005 Commencement

More information

A Binarization Algorithm specialized on Document Images and Photos

A Binarization Algorithm specialized on Document Images and Photos A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a

More information

Lecture 4: Principal components

Lecture 4: Principal components /3/6 Lecture 4: Prncpal components 3..6 Multvarate lnear regresson MLR s optmal for the estmaton data...but poor for handlng collnear data Covarance matrx s not nvertble (large condton number) Robustness

More information

Image Representation & Visualization Basic Imaging Algorithms Shape Representation and Analysis. outline

Image Representation & Visualization Basic Imaging Algorithms Shape Representation and Analysis. outline mage Vsualzaton mage Vsualzaton mage Representaton & Vsualzaton Basc magng Algorthms Shape Representaton and Analyss outlne mage Representaton & Vsualzaton Basc magng Algorthms Shape Representaton and

More information

S1 Note. Basis functions.

S1 Note. Basis functions. S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type

More information

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana

More information

TN348: Openlab Module - Colocalization

TN348: Openlab Module - Colocalization TN348: Openlab Module - Colocalzaton Topc The Colocalzaton module provdes the faclty to vsualze and quantfy colocalzaton between pars of mages. The Colocalzaton wndow contans a prevew of the two mages

More information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

More information

FEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur

FEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur FEATURE EXTRACTION Dr. K.Vjayarekha Assocate Dean School of Electrcal and Electroncs Engneerng SASTRA Unversty, Thanjavur613 41 Jont Intatve of IITs and IISc Funded by MHRD Page 1 of 8 Table of Contents

More information

Range images. Range image registration. Examples of sampling patterns. Range images and range surfaces

Range images. Range image registration. Examples of sampling patterns. Range images and range surfaces Range mages For many structured lght scanners, the range data forms a hghly regular pattern known as a range mage. he samplng pattern s determned by the specfc scanner. Range mage regstraton 1 Examples

More information

Smoothing Spline ANOVA for variable screening

Smoothing Spline ANOVA for variable screening Smoothng Splne ANOVA for varable screenng a useful tool for metamodels tranng and mult-objectve optmzaton L. Rcco, E. Rgon, A. Turco Outlne RSM Introducton Possble couplng Test case MOO MOO wth Game Theory

More information

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation Intellgent Informaton Management, 013, 5, 191-195 Publshed Onlne November 013 (http://www.scrp.org/journal/m) http://dx.do.org/10.36/m.013.5601 Qualty Improvement Algorthm for Tetrahedral Mesh Based on

More information

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision

SLAM Summer School 2006 Practical 2: SLAM using Monocular Vision SLAM Summer School 2006 Practcal 2: SLAM usng Monocular Vson Javer Cvera, Unversty of Zaragoza Andrew J. Davson, Imperal College London J.M.M Montel, Unversty of Zaragoza. josemar@unzar.es, jcvera@unzar.es,

More information

Cluster Analysis of Electrical Behavior

Cluster Analysis of Electrical Behavior Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School

More information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

LECTURE : MANIFOLD LEARNING

LECTURE : MANIFOLD LEARNING LECTURE : MANIFOLD LEARNING Rta Osadchy Some sldes are due to L.Saul, V. C. Raykar, N. Verma Topcs PCA MDS IsoMap LLE EgenMaps Done! Dmensonalty Reducton Data representaton Inputs are real-valued vectors

More information

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning the Kernel Parameters in Kernel Minimum Distance Classifier Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department

More information

Snakes-based approach for extraction of building roof contours from digital aerial images

Snakes-based approach for extraction of building roof contours from digital aerial images Snakes-based approach for extracton of buldng roof contours from dgtal aeral mages Alur P. Dal Poz and Antono J. Fazan São Paulo State Unversty Dept. of Cartography, R. Roberto Smonsen 305 19060-900 Presdente

More information

Polyhedral Surface Smoothing with Simultaneous Mesh Regularization

Polyhedral Surface Smoothing with Simultaneous Mesh Regularization olyhedral Surface Smoothng wth Smultaneous Mesh Regularzaton Yutaka Ohtake The Unversty of Azu Azu-Wakamatsu Cty Fukushma 965-8580 Japan d800@u-azu.ac.jp Alexander G. Belyaev The Unversty of Azu Azu-Wakamatsu

More information

Detection of an Object by using Principal Component Analysis

Detection of an Object by using Principal Component Analysis Detecton of an Object by usng Prncpal Component Analyss 1. G. Nagaven, 2. Dr. T. Sreenvasulu Reddy 1. M.Tech, Department of EEE, SVUCE, Trupath, Inda. 2. Assoc. Professor, Department of ECE, SVUCE, Trupath,

More information

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty

More information

The Codesign Challenge

The Codesign Challenge ECE 4530 Codesgn Challenge Fall 2007 Hardware/Software Codesgn The Codesgn Challenge Objectves In the codesgn challenge, your task s to accelerate a gven software reference mplementaton as fast as possble.

More information

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal

More information

MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION

MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION Paulo Quntlano 1 & Antono Santa-Rosa 1 Federal Polce Department, Brasla, Brazl. E-mals: quntlano.pqs@dpf.gov.br and

More information

Simplification of 3D Meshes

Simplification of 3D Meshes Smplfcaton of 3D Meshes Addy Ngan /4/00 Outlne Motvaton Taxonomy of smplfcaton methods Hoppe et al, Mesh optmzaton Hoppe, Progressve meshes Smplfcaton of 3D Meshes 1 Motvaton Hgh detaled meshes becomng

More information

Modeling, Manipulating, and Visualizing Continuous Volumetric Data: A Novel Spline-based Approach

Modeling, Manipulating, and Visualizing Continuous Volumetric Data: A Novel Spline-based Approach Modelng, Manpulatng, and Vsualzng Contnuous Volumetrc Data: A Novel Splne-based Approach Jng Hua Center for Vsual Computng, Department of Computer Scence SUNY at Stony Brook Talk Outlne Introducton and

More information

Accounting for the Use of Different Length Scale Factors in x, y and z Directions

Accounting for the Use of Different Length Scale Factors in x, y and z Directions 1 Accountng for the Use of Dfferent Length Scale Factors n x, y and z Drectons Taha Soch (taha.soch@kcl.ac.uk) Imagng Scences & Bomedcal Engneerng, Kng s College London, The Rayne Insttute, St Thomas Hosptal,

More information

Computer Animation and Visualisation. Lecture 4. Rigging / Skinning

Computer Animation and Visualisation. Lecture 4. Rigging / Skinning Computer Anmaton and Vsualsaton Lecture 4. Rggng / Sknnng Taku Komura Overvew Sknnng / Rggng Background knowledge Lnear Blendng How to decde weghts? Example-based Method Anatomcal models Sknnng Assume

More information

Reading. 14. Subdivision curves. Recommended:

Reading. 14. Subdivision curves. Recommended: eadng ecommended: Stollntz, Deose, and Salesn. Wavelets for Computer Graphcs: heory and Applcatons, 996, secton 6.-6., A.5. 4. Subdvson curves Note: there s an error n Stollntz, et al., secton A.5. Equaton

More information

Hermite Splines in Lie Groups as Products of Geodesics

Hermite Splines in Lie Groups as Products of Geodesics Hermte Splnes n Le Groups as Products of Geodescs Ethan Eade Updated May 28, 2017 1 Introducton 1.1 Goal Ths document defnes a curve n the Le group G parametrzed by tme and by structural parameters n the

More information

Vectorization of Image Outlines Using Rational Spline and Genetic Algorithm

Vectorization of Image Outlines Using Rational Spline and Genetic Algorithm 01 Internatonal Conference on Image, Vson and Computng (ICIVC 01) IPCSIT vol. 50 (01) (01) IACSIT Press, Sngapore DOI: 10.776/IPCSIT.01.V50.4 Vectorzaton of Image Outlnes Usng Ratonal Splne and Genetc

More information

Real-time Motion Capture System Using One Video Camera Based on Color and Edge Distribution

Real-time Motion Capture System Using One Video Camera Based on Color and Edge Distribution Real-tme Moton Capture System Usng One Vdeo Camera Based on Color and Edge Dstrbuton YOSHIAKI AKAZAWA, YOSHIHIRO OKADA, AND KOICHI NIIJIMA Graduate School of Informaton Scence and Electrcal Engneerng,

More information

Unsupervised Learning

Unsupervised Learning Pattern Recognton Lecture 8 Outlne Introducton Unsupervsed Learnng Parametrc VS Non-Parametrc Approach Mxture of Denstes Maxmum-Lkelhood Estmates Clusterng Prof. Danel Yeung School of Computer Scence and

More information

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like: Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A

More information

Module Management Tool in Software Development Organizations

Module Management Tool in Software Development Organizations Journal of Computer Scence (5): 8-, 7 ISSN 59-66 7 Scence Publcatons Management Tool n Software Development Organzatons Ahmad A. Al-Rababah and Mohammad A. Al-Rababah Faculty of IT, Al-Ahlyyah Amman Unversty,

More information

A high precision collaborative vision measurement of gear chamfering profile

A high precision collaborative vision measurement of gear chamfering profile Internatonal Conference on Advances n Mechancal Engneerng and Industral Informatcs (AMEII 05) A hgh precson collaboratve vson measurement of gear chamferng profle Conglng Zhou, a, Zengpu Xu, b, Chunmng

More information

Chapter 6 Programmng the fnte element method Inow turn to the man subject of ths book: The mplementaton of the fnte element algorthm n computer programs. In order to make my dscusson as straghtforward

More information

Multi-view 3D Position Estimation of Sports Players

Multi-view 3D Position Estimation of Sports Players Mult-vew 3D Poston Estmaton of Sports Players Robbe Vos and Wlle Brnk Appled Mathematcs Department of Mathematcal Scences Unversty of Stellenbosch, South Afrca Emal: vosrobbe@gmal.com Abstract The problem

More information

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning

Outline. Type of Machine Learning. Examples of Application. Unsupervised Learning Outlne Artfcal Intellgence and ts applcatons Lecture 8 Unsupervsed Learnng Professor Danel Yeung danyeung@eee.org Dr. Patrck Chan patrckchan@eee.org South Chna Unversty of Technology, Chna Introducton

More information

Multi-stable Perception. Necker Cube

Multi-stable Perception. Necker Cube Mult-stable Percepton Necker Cube Spnnng dancer lluson, Nobuuk Kaahara Fttng and Algnment Computer Vson Szelsk 6.1 James Has Acknowledgment: Man sldes from Derek Hoem, Lana Lazebnk, and Grauman&Lebe 2008

More information

An Accurate Evaluation of Integrals in Convex and Non convex Polygonal Domain by Twelve Node Quadrilateral Finite Element Method

An Accurate Evaluation of Integrals in Convex and Non convex Polygonal Domain by Twelve Node Quadrilateral Finite Element Method Internatonal Journal of Computatonal and Appled Mathematcs. ISSN 89-4966 Volume, Number (07), pp. 33-4 Research Inda Publcatons http://www.rpublcaton.com An Accurate Evaluaton of Integrals n Convex and

More information

A Robust Method for Estimating the Fundamental Matrix

A Robust Method for Estimating the Fundamental Matrix Proc. VIIth Dgtal Image Computng: Technques and Applcatons, Sun C., Talbot H., Ourseln S. and Adraansen T. (Eds.), 0- Dec. 003, Sydney A Robust Method for Estmatng the Fundamental Matrx C.L. Feng and Y.S.

More information

Fair Triangle Mesh Generation with Discrete Elastica

Fair Triangle Mesh Generation with Discrete Elastica Far Trangle Mesh Generaton wth Dscrete Elastca Shn Yoshzawa, and Alexander G. Belyaev, Computer Graphcs Group, Max-Planck-Insttut für Informatk, 66123 Saarbrücken, Germany Phone: [+49](681)9325-414 Fax:

More information

Fitting & Matching. Lecture 4 Prof. Bregler. Slides from: S. Lazebnik, S. Seitz, M. Pollefeys, A. Effros.

Fitting & Matching. Lecture 4 Prof. Bregler. Slides from: S. Lazebnik, S. Seitz, M. Pollefeys, A. Effros. Fttng & Matchng Lecture 4 Prof. Bregler Sldes from: S. Lazebnk, S. Setz, M. Pollefeys, A. Effros. How do we buld panorama? We need to match (algn) mages Matchng wth Features Detect feature ponts n both

More information

A New Approach For the Ranking of Fuzzy Sets With Different Heights

A New Approach For the Ranking of Fuzzy Sets With Different Heights New pproach For the ankng of Fuzzy Sets Wth Dfferent Heghts Pushpnder Sngh School of Mathematcs Computer pplcatons Thapar Unversty, Patala-7 00 Inda pushpndersnl@gmalcom STCT ankng of fuzzy sets plays

More information

REFRACTION. a. To study the refraction of light from plane surfaces. b. To determine the index of refraction for Acrylic and Water.

REFRACTION. a. To study the refraction of light from plane surfaces. b. To determine the index of refraction for Acrylic and Water. Purpose Theory REFRACTION a. To study the refracton of lght from plane surfaces. b. To determne the ndex of refracton for Acrylc and Water. When a ray of lght passes from one medum nto another one of dfferent

More information

Machine Learning: Algorithms and Applications

Machine Learning: Algorithms and Applications 14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of

More information

Hierarchical clustering for gene expression data analysis

Hierarchical clustering for gene expression data analysis Herarchcal clusterng for gene expresson data analyss Gorgo Valentn e-mal: valentn@ds.unm.t Clusterng of Mcroarray Data. Clusterng of gene expresson profles (rows) => dscovery of co-regulated and functonally

More information

Scan Conversion & Shading

Scan Conversion & Shading Scan Converson & Shadng Thomas Funkhouser Prnceton Unversty C0S 426, Fall 1999 3D Renderng Ppelne (for drect llumnaton) 3D Prmtves 3D Modelng Coordnates Modelng Transformaton 3D World Coordnates Lghtng

More information

Wavefront Reconstructor

Wavefront Reconstructor A Dstrbuted Smplex B-Splne Based Wavefront Reconstructor Coen de Vsser and Mchel Verhaegen 14-12-201212 2012 Delft Unversty of Technology Contents Introducton Wavefront reconstructon usng Smplex B-Splnes

More information

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique //00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy

More information

Backpropagation: In Search of Performance Parameters

Backpropagation: In Search of Performance Parameters Bacpropagaton: In Search of Performance Parameters ANIL KUMAR ENUMULAPALLY, LINGGUO BU, and KHOSROW KAIKHAH, Ph.D. Computer Scence Department Texas State Unversty-San Marcos San Marcos, TX-78666 USA ae049@txstate.edu,

More information

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics

NAG Fortran Library Chapter Introduction. G10 Smoothing in Statistics Introducton G10 NAG Fortran Lbrary Chapter Introducton G10 Smoothng n Statstcs Contents 1 Scope of the Chapter... 2 2 Background to the Problems... 2 2.1 Smoothng Methods... 2 2.2 Smoothng Splnes and Regresson

More information

An Optimal Algorithm for Prufer Codes *

An Optimal Algorithm for Prufer Codes * J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,

More information

Visual Curvature. 1. Introduction. y C. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), June 2007

Visual Curvature. 1. Introduction. y C. IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), June 2007 IEEE onf. on omputer Vson and Pattern Recognton (VPR June 7 Vsual urvature HaRong Lu, Longn Jan Lateck, WenYu Lu, Xang Ba HuaZhong Unversty of Scence and Technology, P.R. hna Temple Unversty, US lhrbss@gmal.com,

More information

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements

2x x l. Module 3: Element Properties Lecture 4: Lagrange and Serendipity Elements Module 3: Element Propertes Lecture : Lagrange and Serendpty Elements 5 In last lecture note, the nterpolaton functons are derved on the bass of assumed polynomal from Pascal s trangle for the fled varable.

More information

Development of an Active Shape Model. Using the Discrete Cosine Transform

Development of an Active Shape Model. Using the Discrete Cosine Transform Development of an Actve Shape Model Usng the Dscrete Cosne Transform Kotaro Yasuda A Thess n The Department of Electrcal and Computer Engneerng Presented n Partal Fulfllment of the Requrements for the

More information

A Bilinear Model for Sparse Coding

A Bilinear Model for Sparse Coding A Blnear Model for Sparse Codng Davd B. Grmes and Rajesh P. N. Rao Department of Computer Scence and Engneerng Unversty of Washngton Seattle, WA 98195-2350, U.S.A. grmes,rao @cs.washngton.edu Abstract

More information

An Image Fusion Approach Based on Segmentation Region

An Image Fusion Approach Based on Segmentation Region Rong Wang, L-Qun Gao, Shu Yang, Yu-Hua Cha, and Yan-Chun Lu An Image Fuson Approach Based On Segmentaton Regon An Image Fuson Approach Based on Segmentaton Regon Rong Wang, L-Qun Gao, Shu Yang 3, Yu-Hua

More information

2-Dimensional Image Representation. Using Beta-Spline

2-Dimensional Image Representation. Using Beta-Spline Appled Mathematcal cences, Vol. 7, 03, no. 9, 4559-4569 HIKARI Ltd, www.m-hkar.com http://dx.do.org/0.988/ams.03.3359 -Dmensonal Image Representaton Usng Beta-plne Norm Abdul Had Faculty of Computer and

More information

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data

Type-2 Fuzzy Non-uniform Rational B-spline Model with Type-2 Fuzzy Data Malaysan Journal of Mathematcal Scences 11(S) Aprl : 35 46 (2017) Specal Issue: The 2nd Internatonal Conference and Workshop on Mathematcal Analyss (ICWOMA 2016) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES

More information

IMPROVING AND EXTENDING THE INFORMATION ON PRINCIPAL COMPONENT ANALYSIS FOR LOCAL NEIGHBORHOODS IN 3D POINT CLOUDS

IMPROVING AND EXTENDING THE INFORMATION ON PRINCIPAL COMPONENT ANALYSIS FOR LOCAL NEIGHBORHOODS IN 3D POINT CLOUDS IMPROVING AND EXTENDING THE INFORMATION ON PRINCIPAL COMPONENT ANALYSIS FOR LOCAL NEIGHBORHOODS IN 3D POINT CLOUDS Davd Belton Cooperatve Research Centre for Spatal Informaton (CRC-SI) The Insttute for

More information

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster

More information

A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION

A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION A NEW FUZZY C-MEANS BASED SEGMENTATION STRATEGY. APPLICATIONS TO LIP REGION IDENTIFICATION Mhaela Gordan *, Constantne Kotropoulos **, Apostolos Georgaks **, Ioanns Ptas ** * Bass of Electroncs Department,

More information

Scan Conversion & Shading

Scan Conversion & Shading 1 3D Renderng Ppelne (for drect llumnaton) 2 Scan Converson & Shadng Adam Fnkelsten Prnceton Unversty C0S 426, Fall 2001 3DPrmtves 3D Modelng Coordnates Modelng Transformaton 3D World Coordnates Lghtng

More information

SIGGRAPH Interactive Image Cutout. Interactive Graph Cut. Interactive Graph Cut. Interactive Graph Cut. Hard Constraints. Lazy Snapping.

SIGGRAPH Interactive Image Cutout. Interactive Graph Cut. Interactive Graph Cut. Interactive Graph Cut. Hard Constraints. Lazy Snapping. SIGGRAPH 004 Interactve Image Cutout Lazy Snappng Yn L Jan Sun Ch-Keung Tang Heung-Yeung Shum Mcrosoft Research Asa Hong Kong Unversty Separate an object from ts background Compose the object on another

More information

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance Tsnghua Unversty at TAC 2009: Summarzng Mult-documents by Informaton Dstance Chong Long, Mnle Huang, Xaoyan Zhu State Key Laboratory of Intellgent Technology and Systems, Tsnghua Natonal Laboratory for

More information

A B-Snake Model Using Statistical and Geometric Information - Applications to Medical Images

A B-Snake Model Using Statistical and Geometric Information - Applications to Medical Images A B-Snake Model Usng Statstcal and Geometrc Informaton - Applcatons to Medcal Images Yue Wang, Eam Khwang Teoh and Dnggang Shen 2 School of Electrcal and Electronc Engneerng, Nanyang Technologcal Unversty

More information

Lecture #15 Lecture Notes

Lecture #15 Lecture Notes Lecture #15 Lecture Notes The ocean water column s very much a 3-D spatal entt and we need to represent that structure n an economcal way to deal wth t n calculatons. We wll dscuss one way to do so, emprcal

More information

Research and Application of Fingerprint Recognition Based on MATLAB

Research and Application of Fingerprint Recognition Based on MATLAB Send Orders for Reprnts to reprnts@benthamscence.ae The Open Automaton and Control Systems Journal, 205, 7, 07-07 Open Access Research and Applcaton of Fngerprnt Recognton Based on MATLAB Nng Lu* Department

More information

Video Object Tracking Based On Extended Active Shape Models With Color Information

Video Object Tracking Based On Extended Active Shape Models With Color Information CGIV'2002: he Frst Frst European Conference Colour on Colour n Graphcs, Imagng, and Vson Vdeo Object rackng Based On Extended Actve Shape Models Wth Color Informaton A. Koschan, S.K. Kang, J.K. Pak, B.

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 256 a course in differential equations for engineering students Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the

More information

Learning Ensemble of Local PDM-based Regressions. Yen Le Computational Biomedicine Lab Advisor: Prof. Ioannis A. Kakadiaris

Learning Ensemble of Local PDM-based Regressions. Yen Le Computational Biomedicine Lab Advisor: Prof. Ioannis A. Kakadiaris Learnng Ensemble of Local PDM-based Regressons Yen Le Computatonal Bomedcne Lab Advsor: Prof. Ioanns A. Kakadars 1 Problem statement Fttng a statstcal shape model (PDM) for mage segmentaton Callosum segmentaton

More information

Biostatistics 615/815

Biostatistics 615/815 The E-M Algorthm Bostatstcs 615/815 Lecture 17 Last Lecture: The Smplex Method General method for optmzaton Makes few assumptons about functon Crawls towards mnmum Some recommendatons Multple startng ponts

More information

TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS. Muradaliyev A.Z.

TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS. Muradaliyev A.Z. TECHNIQUE OF FORMATION HOMOGENEOUS SAMPLE SAME OBJECTS Muradalyev AZ Azerbajan Scentfc-Research and Desgn-Prospectng Insttute of Energetc AZ1012, Ave HZardab-94 E-mal:aydn_murad@yahoocom Importance of

More information

Harmonic Coordinates for Character Articulation PIXAR

Harmonic Coordinates for Character Articulation PIXAR Harmonc Coordnates for Character Artculaton PIXAR Pushkar Josh Mark Meyer Tony DeRose Bran Green Tom Sanock We have a complex source mesh nsde of a smpler cage mesh We want vertex deformatons appled to

More information

A Hierarchical Deformable Model Using Statistical and Geometric Information

A Hierarchical Deformable Model Using Statistical and Geometric Information A Herarchcal Deformable Model Usng Statstcal and Geometrc Informaton Dnggang Shen 3 and Chrstos Davatzkos Department of adology Department of Computer Scence 3 Center for Computer-Integrated Surgcal Systems

More information

A new segmentation algorithm for medical volume image based on K-means clustering

A new segmentation algorithm for medical volume image based on K-means clustering Avalable onlne www.jocpr.com Journal of Chemcal and harmaceutcal Research, 2013, 5(12):113-117 Research Artcle ISSN : 0975-7384 CODEN(USA) : JCRC5 A new segmentaton algorthm for medcal volume mage based

More information

Recognizing Faces. Outline

Recognizing Faces. Outline Recognzng Faces Drk Colbry Outlne Introducton and Motvaton Defnng a feature vector Prncpal Component Analyss Lnear Dscrmnate Analyss !"" #$""% http://www.nfotech.oulu.f/annual/2004 + &'()*) '+)* 2 ! &

More information

An Adaptive-Focus Deformable Model Using Statistical and Geometric Information

An Adaptive-Focus Deformable Model Using Statistical and Geometric Information 906 IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 22, NO. 8, AUGUST 2000 An Adaptve-Focus Deformable Model Usng Statstcal and Geometrc Informaton Dnggang Shen and Chrstos Davatzkos

More information

3D vector computer graphics

3D vector computer graphics 3D vector computer graphcs Paolo Varagnolo: freelance engneer Padova Aprl 2016 Prvate Practce ----------------------------------- 1. Introducton Vector 3D model representaton n computer graphcs requres

More information

3D Virtual Eyeglass Frames Modeling from Multiple Camera Image Data Based on the GFFD Deformation Method

3D Virtual Eyeglass Frames Modeling from Multiple Camera Image Data Based on the GFFD Deformation Method NICOGRAPH Internatonal 2012, pp. 114-119 3D Vrtual Eyeglass Frames Modelng from Multple Camera Image Data Based on the GFFD Deformaton Method Norak Tamura, Somsangouane Sngthemphone and Katsuhro Ktama

More information

Constructing Minimum Connected Dominating Set: Algorithmic approach

Constructing Minimum Connected Dominating Set: Algorithmic approach Constructng Mnmum Connected Domnatng Set: Algorthmc approach G.N. Puroht and Usha Sharma Centre for Mathematcal Scences, Banasthal Unversty, Rajasthan 304022 usha.sharma94@yahoo.com Abstract: Connected

More information

Mesh Editing in ROI with Dual Laplacian

Mesh Editing in ROI with Dual Laplacian Mesh Edtng n ROI wth Dual Laplacan Luo Qong, Lu Bo, Ma Zhan-guo, Zhang Hong-bn College of Computer Scence, Beng Unversty of Technology, Chna lqngng@sohu.com, lubo@but.edu.cn,mzgsy@63.com,zhb@publc.bta.net.cn

More information

A Computer Vision System for Automated Container Code Recognition

A Computer Vision System for Automated Container Code Recognition A Computer Vson System for Automated Contaner Code Recognton Hsn-Chen Chen, Chh-Ka Chen, Fu-Yu Hsu, Yu-San Ln, Yu-Te Wu, Yung-Nen Sun * Abstract Contaner code examnaton s an essental step n the contaner

More information

A Newton-Type Method for Constrained Least-Squares Data-Fitting with Easy-to-Control Rational Curves

A Newton-Type Method for Constrained Least-Squares Data-Fitting with Easy-to-Control Rational Curves A Newton-Type Method for Constraned Least-Squares Data-Fttng wth Easy-to-Control Ratonal Curves G. Cascola a, L. Roman b, a Department of Mathematcs, Unversty of Bologna, P.zza d Porta San Donato 5, 4017

More information

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation

An Iterative Solution Approach to Process Plant Layout using Mixed Integer Optimisation 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 An Iteratve Soluton Approach to Process Plant Layout usng Mxed

More information

An Image Segmentation Method Based on Partial Differential Equation Models

An Image Segmentation Method Based on Partial Differential Equation Models An Image Segmentaton Method Based on Partal Dfferental Equaton Models Jang We, Lu Chan* College of Informaton Engneerng, Tarm Unversty, Alar, Chna *Correspondng Author Emal: 76356718@qq.com Abstract In

More information