Real time 3D face and facial feature tracking

Size: px
Start display at page:

Download "Real time 3D face and facial feature tracking"

Transcription

1 J Real-Time Image Proc (2007) 2:35 44 DOI /s ORIGINAL RESEARCH PAPER Real ime 3D face and facial feaure racking Fadi Dornaika Æ Javier Orozco Received: 23 November 2006 / Acceped: 26 June 2007 / Published online: 28 July 2007 Ó Springer-Verlag 2007 Absrac Deecing and racking human faces in video sequences is useful in a number of applicaions such as gesure recogniion and human-machine ineracion. In his paper, we show ha online appearance models (holisic approaches) can be used for simulaneously racking he head, he lips, he eyebrows, and he eyelids in monocular video sequences. Unlike previous approaches o eyelid racking, we show ha he online appearance models can be used for his purpose. Neiher color informaion nor inensiy edges are used by our proposed approach. More precisely, we show how he classical appearance-based rackers can be upgraded in order o deal wih fas eyelid movemens. The proposed eyelid racking is made robus by avoiding eye feaure exracion. Experimens on real videos show he usefulness of he proposed racking schemes as well as heir enhancemen o our previous approach. Keywords Real-ime racking 3D face racking Facial feaure racking Eyelid racking Online appearance models 1 Inroducion F. Dornaika (&) Insiu Géographique Naional Laboraoire MATIS, 2 Avenue Paseur, Sain Mandé, France fadi.dornaika@ign.fr J. Orozco Compuer Vision Cener, Campus UAB, Bellaerra, Barcelona, Spain orozco@cvc.uab.es Deecing and racking human faces in video sequences is useful in a number of applicaions such as gesure recogniion and human-machine ineracion. Faces play a major role in any human compuer ineracion sysem, because hey represen a rich source of informaion. Faces are he main cue humans use for person deecion/idenificaion. Besides his, faces are he main gaeway o express our feelings and emoional saes. Being able o esimae 3D face pose in real-ime, we can ge a clue abou user s inenions. Vision-based racker sysems provide an aracive alernaive since vision sensors are no invasive. Of paricular ineres are vision-based markerless head and/or face rackers. Since hese rackers do no require any arificial markers o be placed on he face, comforable and naural moions can be achieved. On he oher hand, building robus and real-ime markerless rackers for head and facial feaures is a difficul ask due o he high variabiliy of he face and he facial feaures in videos. To overcome he problem of appearance changes recen works on faces adoped saisical facial exures. For example, he Acive Appearance Models have been proposed as a powerful ool for analyzing facial images [5]. Deerminisic and saisical appearance-based racking mehods have been proposed [2, 4, 12]. These mehods can successfully ackle he image variabiliy and drif problems by using deerminisic or saisical models for he global appearance of a special objec class: he face. A few algorihms exis which aemp o rack boh he head and he facial feaures in real ime, e.g., [2, 12]. These works have addressed he combined head and facial feaure racking using he Acive Appearance Model principles. However, [2, 12] require edious learning sages ha should be performed beforehand and should be repeaed whenever he imaging condiions change. Recenly, we have developed a head and facial feaure racking mehod based on online appearance models (OAMs) [6]. Unlike he acive

2 36 J Real-Time Image Proc (2007) 2:35 44 appearance models, he OAMs offer a lo of flexibiliy and efficiency since hey do no require any facial exure model ha should be compued beforehand. Insead he exure model is buil online from he racked sequence. This paper exends our previous work [6] in wo direcions. Firs, we show ha by adoping a non-occluded shape-free facial exure ha excludes he eye region, more accurae and sable 3D head pose parameers can be obained. Second, unlike feaure-based eyelid rackers, we show ha he OAMs can be used o rack he eyelids. Thus, we can infer he eye sae wihou deecing he eye feaures such as he irises and he eye corners. A shor version of his paper has appeared in [7]. In his paper, we addiionally provide (i) exended experimens, (ii) a descripion abou how o build an appearance-based racker able o deal wih fas eyelid movemens, and (iii) a performance evaluaion aiming a quanifying he eye blink deecion in real condiions. Tracking he eyelids and he irises can be used in many applicaions such as drowsiness deecion and inerfaces for handicapped individuals. For applicaions such as driver awareness sysems, one needs o do more han racking he locaions of he person s eyes bu obain heir deailed descripion. Deecing and racking he eye and is feaures (eye corners, irises, and eyelids) have been addressed by many researchers [11, 14, 16, 19]. However, mos of he proposed approaches rely on inensiy edges and are ime consuming. In [16], deecing he sae of he eye is based on he iris deecion in he sense ha he iris deecion resuls will direcly decide he sae of he eye. In [14], he eyelid sae is inferred from he relaive disance beween he eyelid apex and he iris cener. For each frame in he video, he eyelid conour is deeced using edge pixels and normal flow. The auhors repored ha when he eyes were fully or parially open, he eyelids were successfully locaed and racked 90% of he ime. Their proposed approach depends heavily on he exraced inensiy edges. Moreover, i assumes high-resoluion images depicing an essenially fronal face. In our sudy, we do no use any edges and here is no assumpion on he head pose. In our work, he eyelid moion is inferred a he same ime wih he 3D head pose and oher facial acions, ha is, he eyelid sae does no rely on he deecion resuls of oher feaures such as he eye corners and irises. Tracking he rapid eyelid moion is no a sraighforward ask. In our case, we like o rack he eyelid moion using he principles of OAMs. The challenges are as follows. Firs, he upper eyelid is a highly deformable facial feaure since i has a grea freedom of moion. Second, he eyelid can compleely occlude he iris and sclera, ha is, a facial exure model will have wo differen appearances a he same locaions. Third, he eyelid moion is very fas. The remainder of his paper proceeds as follows. Secion 2 inroduces our deformable 3D facial model as well as he concep of shape-free facial paches. Secion 3 describes he online adapive appearance model. Secion 4 presens a generic racking algorihm ha racks in realime he 3D head pose and some facial acions. I describes how he eyelids are racked using his generic algorihm. Secion 5 gives some comparisons obained wih differen facial paches. In Sec. 6, we presen some racking resuls and one performance evaluaion for eye-blinking deecion. 2 Modeling faces 2.1 A deformable 3D model In our sudy, we use he 3D face model Candide [3]. This 3D deformable wireframe model was firs developed for he purpose of model-based image coding and compuer animaion. The 3D shape of his wireframe model is direcly recorded in coordinae form. I is given by he coordinaes of he 3D verices P i, i =1,, n, where n is he number of verices. Thus, he shape up o a global scale can be fully described by he 3n-vecor g; he concaenaion of he 3D coordinaes of all verices P i. The vecor g is wrien as g ¼ g s þ As a ð1þ where g s is he saic shape of he model, s a he facial acion conrol vecor, and he columns of A are he animaion unis. In his sudy, we use seven modes for he facial animaion unis (AUs) marix A. We have chosen he seven following AUs: lower lip depressor, lip srecher, lip corner depressor, upper lip raiser, eyebrow lowerer, ouer eyebrow raiser and eyelid raiser. These AUs are enough o cover mos common facial animaions. Moreover, hey are essenial for conveying emoions. Thus, he lips are conrolled by four facial acions, he eyebrows are conrolled by wo facial acions, and he eyelids by one facial acion. Figure 1 illusraes hese seven facial acions. In his sudy, Fig. 1 The Candide model and he seven facial acions

3 J Real-Time Image Proc (2007) 2: Fig. 2 The configuraion of he 3D face mesh (fronal view) when each facial acion is se o is maximum value. From lef o righ: lower lip depressor, lip srecher, lip corner depressor, upper lip raiser, eyebrow lowerer, ouer eyebrow raiser he eyelids pair, he inner eyebrows pair, and he ouer eyebrows pair are each conrolled by one facial acion. In Eq. 1, he 3D shape is expressed in a local coordinae sysem. However, one should relae he 3D coordinaes o he image coordinae sysem. To his end, we adop he weak perspecive projecion model. We neglec he perspecive effecs since he deph variaion of he face can be considered as small compared o is absolue deph. Thus, he sae of he 3D wireframe model is given by he 3D head pose parameers (hree roaions and hree ranslaions) and he inernal face animaion conrol vecor s a : This is given by he 13-dimensional vecor b: b ¼½h x ; h y ; h z ; x ; y ; z ; s T a ŠT where: ð2þ h x, h y, and h z represen he hree angles associaed wih he 3D roaion beween he 3D face model coordinae sysem and he camera coordinae sysem. x, y, and z represen he hree componens of he 3D ranslaion vecor beween he 3D face model coordinae sysem and he camera coordinae sysem. Each componen of he vecor s a represens he inensiy of one facial acion. This belongs o he inerval [0, 1] where he zero value corresponds o he neural configuraion (no deformaion) and he one value corresponds o he maximum deformaion. In he sequel, he word facial acion will refer o he facial acion inensiy. Figure 2 illusraes he configuraion of he 3D wireframe model when each facial acion is se o is maximum value. Noe ha if only he aspec raio of he camera is known, hen he componen z (he in-deph ranslaion) is replaced by a scale facor s having he same mapping role beween 3D and 2D. This scale is given by s ¼ a z where a is he focal lengh of he camera in pixels. 2.2 Shape-free facial paches A facial image is represened as a shape-free exure, as shown in Fig. 3b. In his secion, we briefly describe how his shape-free exure is compued from he inpu image and he geomerical parameers b. More deails can be found in [1]. The 2D mesh associaed wih he shape-free exure is obained by projecing he saic shape g s (wireframe), using a cenered fronal 3D pose, ono an image wih a given resoluion. The exure of he shape-free facial image is obained by exure mapping from he riangular 2D mesh covering he face in he inpu image (see Fig. 3a) using a piece-wise affine ransform, W: Similarly o [1], we have aken advanage of he fac ha he barycenric coordinaes of he pixels wihin each riangle are invarian under affine ransforms. In oher words, since he geomery of he 2D mesh in he shape-free image is fixed, he barycenric coordinaes are fixed and can be compued once for all, which considerably reduces he CPU ime associaed wih he exure mapping process he warping process. Once an insance of he 3D model (encoded by he vecor b) is projeced ono he inpu image, he warping process proceeds as follows. The shape-free image bounding he fixed 2D mesh is scanned pixel by pixel. For every scanned pixel in his image, we know is riangle as well as is barycenric coordinaes wihin his riangle. Therefore, he 2D locaion of he corresponding pixel in he inpu image can be easily inferred using a linear combinaion of he coordinaes of he riangle verices where he Fig. 3 a An inpu image wih correc adapaion. b The corresponding shape-free facial pach. c The same pach wihou he eyes region

4 38 J Real-Time Image Proc (2007) 2:35 44 coefficiens are given by he barycenric coordinaes. The greylevel of he scanned pixel is hen se by blending he greylevels associaed wih he four closes pixels o he non-ineger coordinaes of he reurned locaion he bilinear inerpolaion. Mahemaically, he warping process applied o an inpu image y is denoed by xðbþ ¼Wðy; bþ ð3þ where x denoes he shape-free facial exure and b denoes he geomerical parameers. Wihou loss of generaliy, we have used wo resoluion levels for he shape-free exures, encoded by 1,310 and 5,392 facial pixels bounded by and recangular boxes, respecively. Obviously, oher levels can be used. Generally speaking as he resoluion increases, he racking accuracy increases. However, by experience we found ha he second resoluion level is a good rade-off beween he accuracy and he compuaional cos. To parially compensae for conras variaions, he original shape-free exure x is ransformed ino an image having a mean equal o 0 and a variance equal o 1. The complee image ransformaion is implemened as follows: (i) ransfer he exure y using he piece-wise affine ransform associaed wih he geomeric parameers b ¼½h x ; h y ; h z ; x ; y ; z ; s T a ŠT ; and (ii) perform zero-meanuni-variance normalizaion on he obained pach. Figure 3 illusraes wo shape-free paches associaed wih an inpu image. 3 Problem formulaion and adapive appearance models Given a video sequence depicing a moving head/face, we would like o recover, for each frame, he 3D head pose and he facial acions encoded by he conrol vecor s a : In oher words, we would like o esimae he vecor b (2) a ime given all he observed daa unil ime, denoed y 1: {y 1,, y }. In a racking conex, he model parameers associaed wih he curren frame will be handed over o he nex frame. For each inpu frame y, he observaion is he shape-free facial pach associaed wih he geomeric parameers b. We use he HAT symbol for he racked parameers and paches. For a given frame ; ^b represens he compued geomeric parameers and ^x he corresponding shape-free pach, ha is, ^x ¼ xð^b Þ¼Wðy ; ^b Þ ð4þ The esimaion of he curren parameers ^b from he previous ones ^b 1 and from he sequence of images will be presened in Sec. 4. In our work, he iniial parameers ^b 1 corresponding o he firs frame are manually provided. The auomaic iniializaion can be obained using he saisical echnique proposed in [2]. By assuming ha he pixels wihin he shape-free pach are independen, we can model he appearance of he shape-free facial pach using a mulivariae Gaussian wih a diagonal covariance marix R: Le l be he Gaussian cener and r be he vecor conaining he square roo of he diagonal elemens of he covariance marix R: l and r are d-vecors (d is he size of x) represening he appearance parameers. In summary, he observaion likelihood a ime is wrien as pðy jb Þ¼pðx jb Þ¼ Yd i¼1 Nðx i ; l i ; r i Þ where N(x i ; l i,r i ) is a normal densiy: " Nðx i ; l i ; r i Þ¼ð2pr 2 i Þ 1=2 exp 1 # x i l 2 i 2 r i ð5þ ð6þ We assume ha he appearance model summarizes he pas observaions under an exponenial envelope, ha is, he pas observaions are exponenially forgoen. When he appearance is racked for he curren inpu image, i.e. he exure ^x is available, we can updae he appearance and use i o rack in he nex frame. I can be shown ha he appearance model parameers, i.e., l and r can be updaed using he following equaions (see [9] for more deails on online appearance models): l iðþ1þ ¼ð1 aþl iðþ þ a^x iðþ ð7þ r 2 i ðþ1þ ¼ð1 aþr 2 i ðþ þ að^x iðþ l iðþ Þ 2 ð8þ In he above equaions, he subscrip i denoes a pixel in he pach ^x: This echnique, also called recursive filering, is simple, ime-efficien and herefore suiable for real-ime applicaions. The appearance parameers reflec he mos recen observaions wihin a roughly L = 1/a window wih exponenial decay. Noe ha l is iniialized wih he firs pach ^x 1 corresponding o he geomerical parameers ^b 1 : However, Eq. 8 is no used unil he number of frames reaches a given value (e.g., he firs 40 frames). For hese frames, he classical variance is used, ha is, Eq. 8 is used wih a being se o 1=: Here, we used a single Gaussian o model he appearance of each pixel in he shape-free pach. However, modeling he appearance wih Gaussian mixures can also be used on he expense of some addiional compuaional load (e.g., see [10, 18]).

5 J Real-Time Image Proc (2007) 2: Tracking using adapive appearance regisraion 4.1 A generic racking algorihm In his secion, we describe he racking algorihm ha is used for racking he head and he facial feaures in a monocular video sequence. The racked facial acions as well as he shape-free facial pach can be arbirary. We consider he sae vecor b ¼½h x ; h y ; h z ; x ; y ; z ; s T a ŠT encapsulaing he 3D head pose parameers and he facial acions. If we only rack he lips and he eyebrows hen he vecor s a will encode six facial acions, i.e., he vecor b will have 12 componens. If we rack he lips, he eyebrows, and he eyelids hen s a will encode seven facial acions, i.e., he vecor b will have 13 componens. We will show how his sae vecor can be recovered for ime using (i) he previous known sae ^b 1 ; (ii) he curren inpu image y, and (iii) he curren appearance parameers encoded by he mean l and he covariance marix R: The sough parameers b a ime are esimaed by regisering he inpu image (warped version) o he curren appearance model. For his purpose, we minimize he Mahalanobis disance beween he warped exure and he curren appearance mean, min b eðb Þ¼min Dðxðb Þ; l Þ¼ Xd b i¼1 x i l 2 i ð9þ r i The above crierion can be minimized using ieraive firs-order linear approximaion which is equivalen o a Gauss-Newon mehod. I is worhwhile noing ha minimizing he above crierion is equivalen o maximizing he likelihood measure given by (5). Moreover, he above opimizaion is made robus by using robus saisics [6]. For every frame in he video sequence, he corresponding sae vecor b is esimaed using he following. The whole racking algorihm is oulined in Fig Tracking algorihm Saring from b ¼ ^b 1 ; we compue he error vecor Wðy ; ^b 1 Þ l ¼ x ð^b 1 Þ l and he corresponding Mahalanobis disance e(b) (given by Eq. 9). We find a shif Db by muliplying he error vecor wih he negaive pseudo-inverse of he gradien marix G ¼ ox ob using (10). Db ¼ G y ðwðy ; ^b 1 Þ l Þ ð10þ where G =(G T G ) 1 G T is he pseudo-inverse of G. The vecor Db gives a displacemen in he search space for which he error, e, can be minimized. We compue a new parameer vecor and a new error: b 0 ¼ b þ qdb e 0 ¼ eðb 0 Þ ð11þ Fig. 4 The generic appearancebased racking algorihm. For every image in he monocular video sequence, he 3D head pose parameers as well as he facial acions are simulaneously esimaed by regisering he curren exure wih he curren facial exure model he curren appearance model ) ( D( ( ), ) ( ( ) y, x, y, z, x, z, T a 1 ( ) s a ^ ^ +1 = (1 - ) = (1 - ) 2 + (x - ) 2

6 40 J Real-Time Image Proc (2007) 2:35 44 where q is a posiive real. If e < e, we updae b according o (11) and he process is ieraed unil convergence. If e e, we ry smaller updae seps in he same direcion (i.e., a smaller q is used). Convergence is declared when he error canno be improved anymore or he number of ieraions reaches a maximum. In he above opimizaion, he gradien marix G ¼ owðy ;b Þ ob ¼ ox ob is approximaed by numerical differences similar o he work of Cooes [5]. Noice ha he gradien marix is compued for each ime sep. The advanage is wofold. Firs, a varying gradien marix is able o accommodae appearance changes. Second, i will be closer o he exac gradien marix since i is compued for he curren geomeric configuraion (3D head pose and facial acions) whereas a fixed gradien marix can be a source of errors. More deails abou his opimizaion echnique can be found in [6]. 4.2 Eyelids racking As we have menioned earlier, racking he eyelid moion is a challenging ask, and mos of he proposed approaches for locaing and racking he eyelids rely on he exraced inensiy edges. In our case, he generic racking algorihm (Sec. 4.1) is used for boh cases: (i) racking he lips and he eyebrows, and (ii) racking he lips, he eyebrows, and he eyelids. However, in he second case, here are wo main differences. Firs, we adop a shape-free facial pach whose eye region corresponds o closed eyes configuraion (see Fig. 5), which excludes he iris and sclera regions. Noe ha he 2D shape of he eyelids in he shape-free pach is fixed like any oher facial feaure, ha is, he eyelids appear closed in he facial pach regardless of he sae of he eyelids in he inpu image. This is illusraed in Fig. 5. The maximum value of he eyelid facial acion corresponds o wide open eyes while he zero value corresponds o closed eyes. Noe ha when he eyes are open in he inpu image, he exure of he eyelid region in he shape-free pach (associaed wih a correc eyelid facial acion) will be a disored version of a very small area in he inpu image. However, he global The inpu image The 2D shape The facial pach Fig. 5 The shape-free facial pach used o rack 13 degrees of freedom including he eyelid moion appearance of he eyelid is sill preserved since he eyelids have he skin appearance. Second, since he eyelid moion is very fas a good esimaion of is gradien (a column in he global gradien marix G ¼ ox =ob) is compued wih a large number of perurbaion seps ha cover almos all he variaion inerval. 5 Tracking comparisons In his secion, we compare he 3D head pose esimaes obained wih wo differen shape-free facial paches using he same racking algorihm described above (Sec. 4.1) and he same sae vecor b given by he six head pose parameers and he six facial acions associaed wih he lips and eyebrows, ha is, he eyelid facial acion is no used. To his end, we use he wo shape-free paches depiced in Fig. 3. The firs pach includes a region for he eye feaures namely he iris and he sclera. The second pach is obained from he firs one by only removing he eyes region. We have used a 1,000-frame sequence feauring a alking subjec 1 as a es video. Noe ha alking is a sponaneous aciviy. Figure 6 illusraes he esimaed 3D head pose parameers associaed wih a 150-frame segmen using he wo differen shape-free facial paches. The displayed parameers are (from op o boom): he pich angle h x, he yaw angle h y, he scale s, and he verical ranslaion y. The video segmen sars a frame 500 and conains hree eye blinks a frames 10, 104, and 145. The solid curves correspond o he firs facial pach (wih eye region) while he doed curves correspond o he second pach (wihou eye region). One can noice ha (i) he solid curves and he doed curves coincide for almos all frames, and (ii) he mos significan discrepancies occur a hose frames associaed wih an eye blink (e.g., see he scale plo). Alhough here is no ground-ruh daa for he head moions, we found ha hese discrepancies correspond well o some inroduced errors since acually he head has no suddenly moved a hese ime insans (see frames 104 and 145). Thus, o some exen hese discrepancies can be considered as errors associaed wih he esimaed parameers. Whenever an eye blink occurs he pach wihou he eyes region has provided more accurae and sable parameers han he pach wih he eye region. This is explained by he fac ha he pach conaining he eye region (sclera and iris) does no model he shape of he eyelids. Therefore, despie he use of robus saisics he esimaion 1 hp://

7 J Real-Time Image Proc (2007) 2: Pich angle 4 Wih Eyes Region Wihou Eyes Region 2 0 Deg Frames 6 7 Yaw angle Wih Eyes Region Wihou Eyes Region 8 Deg Frames Scale Wih Eyes Region Wihou Eyes Region Frames 0.8 Open Eyelid moion Closed Frames Fig. 7 Tracking he 3D head pose, he lips, he eyebrows and he eyelids associaed wih a 1,000-frame sequence. Only frames 280, 284, 975 are shown. The plo depics he esimaed eyelid facial acion as a funcion of he sequence frames Pixels Y Translaion Wih Eyes Region Wihou Eyes Region of he 3D head pose parameers wih a pach conaining he eye region (sclera and iris) is affeced by he eyelid moion. One can noice ha he roaional discrepancies seem o be small (abou one degree). However, he verical and indeph ranslaion errors can be large. For example, a frame 145 he obained scale discrepancy is abou 0.025, which corresponds o an in-deph error of abou 3 cm Frames Fig. 6 3D head pose parameers obained wih wo differen facial paches ha differ by he eyes region 2 The exac value depends on he camera-inrinsic parameers and he absolue deph.

8 42 J Real-Time Image Proc (2007) 2:35 44 Fig. 9 A frame belonging o he low resoluion video depiced a he op of Fig. 8 zoomed views of hose frames. Noice how he eyelids are correcly racked in he inpu images. The esimaed eyelid facial acion reflecs he degree of he eye openness. The upper lef corner of each image shows he curren appearance ðl Þ and he curren shape-free exure ð^x Þ: The boom of his figure displays he esimaed eyelid facial acion as a funcion of he sequence frames where zero value corresponds o a closed eye and one value o a wide open eye. Figure 8 displays some racking resuls obained wih four video sequences. The firs video has a low resoluion. Figure 9 displays a snapsho of he low resoluion video. 6.1 Eye blink deecion Fig. 8 Tracking resuls associaed wih four video sequences. The firs one has a low resoluion 6 Head, lips, eyebrows, and eyelid racking In he previous secion, we have shown ha he accuracy of he racked 3D head pose parameers can be affeced by he eyelid moion (eye blinking) if he sclera and iris region is included in he exure model. This is no surprising since eye blinking corresponds o a sudden occlusion of a small par of he face. Thus, if he eyelid moion is racked one can expec ha he 3D head pose parameers can be more sable. We have racked he head, lips, eyebrows, and eyelids using he 1,000-frame sequence. Figure 7 displays he racking resuls (13 degrees of freedom) associaed wih frames 280, 284, and 975. The lef column displays Eye blinking is a discree and imporan facial acion [8, 13, 15]. The rae of blinking varies, bu on average he eye blinks once every 5 s. 3 In our case, he eye-blinks can be direcly deeced and segmened by hresholding he racked eyelid facial acion. As can be seen, he dual sae of he eye can easily be inferred from he coninuous curve associaed wih he eyelid facial acion. For he racked sequences, all eye blinks are correcly deeced and segmened. Figure 10 illusraes he racking resuls associaed wih a 594-frame sequence. This sequence depics a subjec performing head moions and facial animaions. This sequence depics 18 eye blinks. The proposed algorihm was able o correcly deec 17 blinks. The non-deeced blink happened a he same ime when he subjec pu on his glasses, ha is, he eyelids are suddenly occluded by he frame of he eyeglasses, which corresponds o a sudden appearance variaion. However, once he glasses are pu on all subsequen eye blinks are correcly deeced. 3 hp://

9 J Real-Time Image Proc (2007) 2: akes 333 ms on a Penium II 400 MHz PC. This mehod only racks he iris locaion as well as he eye sae. In [17], he average running ime for racking he eyelids and eye corners in one frame is abou 100 ms. However, his mehod only racks he eye corners and he eyelids. 7 Conclusion In his paper, we have exended our appearance-based 3D head and facial acion racker o deal wih eyelid moions. The 3D head pose and he facial acions associaed wih he lips, eyebrows, and eyelids are simulaneously esimaed in real-ime using OAMs. Compared wih oher eyelid racking echniques our proposed approach has several advanages. Firs, compuing and segmening inensiy edges has been avoided. Second, he eyelid racking does no depend on he deecion of oher eye feaures. Third, he eyelid moion is racked using a coninuous facial acion. Experimens on real video sequences including low-resoluion videos indicae ha he eye sae can be deeced using he eyelid racking resuls. Acknowledgmens The auhors hank Dr. Franck Davoine from CNRS, Compiegne, France, for providing he video sequence shown in Figure 10. References Fig. 10 Tracking resuls associaed wih one video sequence depicing 18 eye blinks Processing ime On a 3.2-GHz PC, a non-opimized C code of our proposed approach compues he 3D head pose and he seven facial acions in 70 ms. The edge-based mehod presened in [16] 1. Ahlberg, J.: Real-ime facial feaure racking using an acive model wih fas image warping. In: Inernaional Workshop on Very Low Birae Video (VLBV). Ahens, Greece (2001) 2. Ahlberg, J.: An acive model for facial feaure racking. EURA- SIP J. Appl. Signal Process. 2002(6), (2002) 3. Ahlberg, J: Model-based coding: exracion, coding, and evaluaion of face model parameers. Ph.D. hesis, No. 761, Linköping Universiy, Sweden (2002) 4. Cascia, M., Sclaroff, S., Ahisos, V.: Fas, reliable head racking under varying illuminaion: an approach based on regisraion of exure-mapped 3D models. IEEE Trans. Paern Anal. Mach. Inell. 22(4), (2000) 5. Cooes, T., Edwards, G., Taylor, C.: Acive appearance models. IEEE Trans. Paern Anal. Mach. Inell. 23(6): (2001) 6. Dornaika, F., Davoine, F.: On appearance based face and facial acion racking. IEEE Trans. Circuis Sys. Video Technol. 16(9): (2006) 7. Dornaika, F., Orozco, J., Gonzalez, J.: Combined head, lips, eyebrows, and eyelids racking using adapive appearance models. In: LNCS IV Conference on Ariculaed Moion and Deformable Objecs, pp (2006) 8. Grauman, K., Beke, M., Gips, J., Bradski, G.R.: Communicaion via eye blinks: deecion and duraion analysis in real ime. In: Inernaional Conference on Compuer Vision and Paern Recogniion (2001) 9. Jepson, A., Flee, D., El-Maraghi, T. Robus online appearance models for visual racking. IEEE Trans. Paern Anal. Mach. Inell. 25(10): (2003) 10. Lee, D.: Effecive Gaussian mixure learning for video background subracion. IEEE Trans. Paern Anal. Mach. Inell. 27(5): (2005)

10 44 J Real-Time Image Proc (2007) 2: Liu, H., Wu, Y., Zha, H.: Eye saes deecion from color facial image sequence. In: SPIE inernaional conference on image and graphics, vol. 4875, pp (2002) 12. Mahews, I., Baker, S.: Acive appearance models revisied. In. J. Compu. Vis. 60(2): (2004) 13. Moriyama, T., Kanade, T., Cohn, J., Xiao, J., Ambadar, Z., Gao, J., Imamura, H.: Auomaic recogniion of eye blinking in sponaneously occuring behavior. In: Inernaional Conference on Paern Recogniion (2002) 14. Sirohey, S., Rosenfeld, A., Duric, Z.: A mehod of deecing and racking irises and eyelids in video. Paern Recogni. 35(6): (2002) 15. Tan, H., Zhang, Y.J.: Deecing eye blink saes by racking iris and eyelids. Paern Recogni. Le. 27(6): (2006) 16. Tian, Y., Kanade, T., Cohn, J.F.: Dual-sae parameric eye racking. In: Inernaional Conference on Auomaic Face and Gesure Recogniion (2000) 17. Uzunova, V.I.: An eyelids and eye corners deecion and racking mehod for rapid iris racking. Maser s hesis, Universiy of Magdeburg, Zhou, S., Chellappa, R., Mogghaddam, B.: Visual racking and recogniion using appearance-adapive models in paricle filers. IEEE Trans. Image Process. 13(11): (2004) 19. Zhu, J., Yang, J.: Subpixel eye gaze racking. In: Inernaional Conference on Auomaic Face and Gesure Recogniion (2002)

Implementing Ray Casting in Tetrahedral Meshes with Programmable Graphics Hardware (Technical Report)

Implementing Ray Casting in Tetrahedral Meshes with Programmable Graphics Hardware (Technical Report) Implemening Ray Casing in Terahedral Meshes wih Programmable Graphics Hardware (Technical Repor) Marin Kraus, Thomas Erl March 28, 2002 1 Inroducion Alhough cell-projecion, e.g., [3, 2], and resampling,

More information

CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL

CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL Klečka Jan Docoral Degree Programme (1), FEEC BUT E-mail: xkleck01@sud.feec.vubr.cz Supervised by: Horák Karel E-mail: horak@feec.vubr.cz

More information

STEREO PLANE MATCHING TECHNIQUE

STEREO PLANE MATCHING TECHNIQUE STEREO PLANE MATCHING TECHNIQUE Commission III KEY WORDS: Sereo Maching, Surface Modeling, Projecive Transformaion, Homography ABSTRACT: This paper presens a new ype of sereo maching algorihm called Sereo

More information

Visual Perception as Bayesian Inference. David J Fleet. University of Toronto

Visual Perception as Bayesian Inference. David J Fleet. University of Toronto Visual Percepion as Bayesian Inference David J Flee Universiy of Torono Basic rules of probabiliy sum rule (for muually exclusive a ): produc rule (condiioning): independence (def n ): Bayes rule: marginalizaion:

More information

FACIAL ACTION TRACKING USING PARTICLE FILTERS AND ACTIVE APPEARANCE MODELS. Soumya Hamlaoui & Franck Davoine

FACIAL ACTION TRACKING USING PARTICLE FILTERS AND ACTIVE APPEARANCE MODELS. Soumya Hamlaoui & Franck Davoine FACIAL ACTION TRACKING USING PARTICLE FILTERS AND ACTIVE APPEARANCE MODELS Soumya Hamlaoui & Franck Davoine HEUDIASYC Mixed Research Uni, CNRS / Compiègne Universiy of Technology BP 20529, 60205 Compiègne

More information

Evaluation and Improvement of Region-based Motion Segmentation

Evaluation and Improvement of Region-based Motion Segmentation Evaluaion and Improvemen of Region-based Moion Segmenaion Mark Ross Universiy Koblenz-Landau, Insiue of Compuaional Visualisics, Universiässraße 1, 56070 Koblenz, Germany Email: ross@uni-koblenz.de Absrac

More information

EECS 487: Interactive Computer Graphics

EECS 487: Interactive Computer Graphics EECS 487: Ineracive Compuer Graphics Lecure 7: B-splines curves Raional Bézier and NURBS Cubic Splines A represenaion of cubic spline consiss of: four conrol poins (why four?) hese are compleely user specified

More information

Research Article Auto Coloring with Enhanced Character Registration

Research Article Auto Coloring with Enhanced Character Registration Compuer Games Technology Volume 2008, Aricle ID 35398, 7 pages doi:0.55/2008/35398 Research Aricle Auo Coloring wih Enhanced Characer Regisraion Jie Qiu, Hock Soon Seah, Feng Tian, Quan Chen, Zhongke Wu,

More information

A Fast Stereo-Based Multi-Person Tracking using an Approximated Likelihood Map for Overlapping Silhouette Templates

A Fast Stereo-Based Multi-Person Tracking using an Approximated Likelihood Map for Overlapping Silhouette Templates A Fas Sereo-Based Muli-Person Tracking using an Approximaed Likelihood Map for Overlapping Silhouee Templaes Junji Saake Jun Miura Deparmen of Compuer Science and Engineering Toyohashi Universiy of Technology

More information

Learning in Games via Opponent Strategy Estimation and Policy Search

Learning in Games via Opponent Strategy Estimation and Policy Search Learning in Games via Opponen Sraegy Esimaion and Policy Search Yavar Naddaf Deparmen of Compuer Science Universiy of Briish Columbia Vancouver, BC yavar@naddaf.name Nando de Freias (Supervisor) Deparmen

More information

MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES

MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES B. MARCOTEGUI and F. MEYER Ecole des Mines de Paris, Cenre de Morphologie Mahémaique, 35, rue Sain-Honoré, F 77305 Fonainebleau Cedex, France Absrac. In image

More information

Robust parameterized component analysis: theory and applications to 2D facial appearance models

Robust parameterized component analysis: theory and applications to 2D facial appearance models Compuer Vision and Image Undersanding 91 (2003) 53 71 www.elsevier.com/locae/cviu Robus parameerized componen analysis: heory and applicaions o 2D facial appearance models Fernando De la Torre a, * and

More information

Improved TLD Algorithm for Face Tracking

Improved TLD Algorithm for Face Tracking Absrac Improved TLD Algorihm for Face Tracking Huimin Li a, Chaojing Yu b and Jing Chen c Chongqing Universiy of Poss and Telecommunicaions, Chongqing 400065, China a li.huimin666@163.com, b 15023299065@163.com,

More information

A Face Detection Method Based on Skin Color Model

A Face Detection Method Based on Skin Color Model A Face Deecion Mehod Based on Skin Color Model Dazhi Zhang Boying Wu Jiebao Sun Qinglei Liao Deparmen of Mahemaics Harbin Insiue of Technology Harbin China 150000 Zhang_dz@163.com mahwby@hi.edu.cn sunjiebao@om.com

More information

4.1 3D GEOMETRIC TRANSFORMATIONS

4.1 3D GEOMETRIC TRANSFORMATIONS MODULE IV MCA - 3 COMPUTER GRAPHICS ADMN 29- Dep. of Compuer Science And Applicaions, SJCET, Palai 94 4. 3D GEOMETRIC TRANSFORMATIONS Mehods for geomeric ransformaions and objec modeling in hree dimensions

More information

A METHOD OF MODELING DEFORMATION OF AN OBJECT EMPLOYING SURROUNDING VIDEO CAMERAS

A METHOD OF MODELING DEFORMATION OF AN OBJECT EMPLOYING SURROUNDING VIDEO CAMERAS A METHOD OF MODELING DEFORMATION OF AN OBJECT EMLOYING SURROUNDING IDEO CAMERAS Joo Kooi TAN, Seiji ISHIKAWA Deparmen of Mechanical and Conrol Engineering Kushu Insiue of Technolog, Japan ehelan@is.cnl.kuech.ac.jp,

More information

CENG 477 Introduction to Computer Graphics. Modeling Transformations

CENG 477 Introduction to Computer Graphics. Modeling Transformations CENG 477 Inroducion o Compuer Graphics Modeling Transformaions Modeling Transformaions Model coordinaes o World coordinaes: Model coordinaes: All shapes wih heir local coordinaes and sies. world World

More information

Wheelchair-user Detection Combined with Parts-based Tracking

Wheelchair-user Detection Combined with Parts-based Tracking Wheelchair-user Deecion Combined wih Pars-based Tracking Ukyo Tanikawa 1, Yasuomo Kawanishi 1, Daisuke Deguchi 2,IchiroIde 1, Hiroshi Murase 1 and Ryo Kawai 3 1 Graduae School of Informaion Science, Nagoya

More information

Design Alternatives for a Thin Lens Spatial Integrator Array

Design Alternatives for a Thin Lens Spatial Integrator Array Egyp. J. Solids, Vol. (7), No. (), (004) 75 Design Alernaives for a Thin Lens Spaial Inegraor Array Hala Kamal *, Daniel V azquez and Javier Alda and E. Bernabeu Opics Deparmen. Universiy Compluense of

More information

Real Time Integral-Based Structural Health Monitoring

Real Time Integral-Based Structural Health Monitoring Real Time Inegral-Based Srucural Healh Monioring The nd Inernaional Conference on Sensing Technology ICST 7 J. G. Chase, I. Singh-Leve, C. E. Hann, X. Chen Deparmen of Mechanical Engineering, Universiy

More information

In Proceedings of CVPR '96. Structure and Motion of Curved 3D Objects from. using these methods [12].

In Proceedings of CVPR '96. Structure and Motion of Curved 3D Objects from. using these methods [12]. In Proceedings of CVPR '96 Srucure and Moion of Curved 3D Objecs from Monocular Silhouees B Vijayakumar David J Kriegman Dep of Elecrical Engineering Yale Universiy New Haven, CT 652-8267 Jean Ponce Compuer

More information

PART 1 REFERENCE INFORMATION CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONITOR

PART 1 REFERENCE INFORMATION CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONITOR . ~ PART 1 c 0 \,).,,.,, REFERENCE NFORMATON CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONTOR n CONTROL DATA 6400 Compuer Sysems, sysem funcions are normally handled by he Monior locaed in a Peripheral

More information

Image segmentation. Motivation. Objective. Definitions. A classification of segmentation techniques. Assumptions for thresholding

Image segmentation. Motivation. Objective. Definitions. A classification of segmentation techniques. Assumptions for thresholding Moivaion Image segmenaion Which pixels belong o he same objec in an image/video sequence? (spaial segmenaion) Which frames belong o he same video sho? (emporal segmenaion) Which frames belong o he same

More information

Video-Based Face Recognition Using Probabilistic Appearance Manifolds

Video-Based Face Recognition Using Probabilistic Appearance Manifolds Video-Based Face Recogniion Using Probabilisic Appearance Manifolds Kuang-Chih Lee Jeffrey Ho Ming-Hsuan Yang David Kriegman klee10@uiuc.edu jho@cs.ucsd.edu myang@honda-ri.com kriegman@cs.ucsd.edu Compuer

More information

DAGM 2011 Tutorial on Convex Optimization for Computer Vision

DAGM 2011 Tutorial on Convex Optimization for Computer Vision DAGM 2011 Tuorial on Convex Opimizaion for Compuer Vision Par 3: Convex Soluions for Sereo and Opical Flow Daniel Cremers Compuer Vision Group Technical Universiy of Munich Graz Universiy of Technology

More information

J. Vis. Commun. Image R.

J. Vis. Commun. Image R. J. Vis. Commun. Image R. 20 (2009) 9 27 Conens liss available a ScienceDirec J. Vis. Commun. Image R. journal homepage: www.elsevier.com/locae/jvci Face deecion and racking using a Boosed Adapive Paricle

More information

AUTOMATIC 3D FACE REGISTRATION WITHOUT INITIALIZATION

AUTOMATIC 3D FACE REGISTRATION WITHOUT INITIALIZATION Chaper 3 AUTOMATIC 3D FACE REGISTRATION WITHOUT INITIALIZATION A. Koschan, V. R. Ayyagari, F. Boughorbel, and M. A. Abidi Imaging, Roboics, and Inelligen Sysems Laboraory, The Universiy of Tennessee, 334

More information

NEWTON S SECOND LAW OF MOTION

NEWTON S SECOND LAW OF MOTION Course and Secion Dae Names NEWTON S SECOND LAW OF MOTION The acceleraion of an objec is defined as he rae of change of elociy. If he elociy changes by an amoun in a ime, hen he aerage acceleraion during

More information

THE goal of this work is to develop statistical models for

THE goal of this work is to develop statistical models for IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 32, NO. 4, APRIL 2010 579 Nonsaionary Shape Aciviies: Dynamic Models for Landmark Shape Change and Applicaions Samarji Das, Suden Member,

More information

Occlusion-Free Hand Motion Tracking by Multiple Cameras and Particle Filtering with Prediction

Occlusion-Free Hand Motion Tracking by Multiple Cameras and Particle Filtering with Prediction 58 IJCSNS Inernaional Journal of Compuer Science and Nework Securiy, VOL.6 No.10, Ocober 006 Occlusion-Free Hand Moion Tracking by Muliple Cameras and Paricle Filering wih Predicion Makoo Kao, and Gang

More information

IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS PART A: SYSTEMS AND HUMANS 1

IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS PART A: SYSTEMS AND HUMANS 1 TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS PART A: SYSTEMS AND HUMANS 1 Adapive Appearance Model and Condensaion Algorihm for Robus Face Tracking Yui Man Lui, Suden Member,, J. Ross Beveridge, Member,,

More information

A Matching Algorithm for Content-Based Image Retrieval

A Matching Algorithm for Content-Based Image Retrieval A Maching Algorihm for Conen-Based Image Rerieval Sue J. Cho Deparmen of Compuer Science Seoul Naional Universiy Seoul, Korea Absrac Conen-based image rerieval sysem rerieves an image from a daabase using

More information

The Impact of Product Development on the Lifecycle of Defects

The Impact of Product Development on the Lifecycle of Defects The Impac of Produc Developmen on he Lifecycle of Rudolf Ramler Sofware Compeence Cener Hagenberg Sofware Park 21 A-4232 Hagenberg, Ausria +43 7236 3343 872 rudolf.ramler@scch.a ABSTRACT This paper invesigaes

More information

Projection & Interaction

Projection & Interaction Projecion & Ineracion Algebra of projecion Canonical viewing volume rackball inerface ransform Hierarchies Preview of Assignmen #2 Lecure 8 Comp 236 Spring 25 Projecions Our lives are grealy simplified

More information

Real-time 2D Video/3D LiDAR Registration

Real-time 2D Video/3D LiDAR Registration Real-ime 2D Video/3D LiDAR Regisraion C. Bodenseiner Fraunhofer IOSB chrisoph.bodenseiner@iosb.fraunhofer.de M. Arens Fraunhofer IOSB michael.arens@iosb.fraunhofer.de Absrac Progress in LiDAR scanning

More information

Robust Visual Tracking for Multiple Targets

Robust Visual Tracking for Multiple Targets Robus Visual Tracking for Muliple Targes Yizheng Cai, Nando de Freias, and James J. Lile Universiy of Briish Columbia, Vancouver, B.C., Canada, V6T 1Z4 {yizhengc, nando, lile}@cs.ubc.ca Absrac. We address

More information

A Bayesian Approach to Video Object Segmentation via Merging 3D Watershed Volumes

A Bayesian Approach to Video Object Segmentation via Merging 3D Watershed Volumes A Bayesian Approach o Video Objec Segmenaion via Merging 3D Waershed Volumes Yu-Pao Tsai 1,3, Chih-Chuan Lai 1,2, Yi-Ping Hung 1,2, and Zen-Chung Shih 3 1 Insiue of Informaion Science, Academia Sinica,

More information

Probabilistic Detection and Tracking of Motion Discontinuities

Probabilistic Detection and Tracking of Motion Discontinuities Probabilisic Deecion and Tracking of Moion Disconinuiies Michael J. Black David J. Flee Xerox Palo Alo Research Cener 3333 Coyoe Hill Road Palo Alo, CA 94304 fblack,fleeg@parc.xerox.com hp://www.parc.xerox.com/fblack,fleeg/

More information

Real-Time Non-Rigid Multi-Frame Depth Video Super-Resolution

Real-Time Non-Rigid Multi-Frame Depth Video Super-Resolution Real-Time Non-Rigid Muli-Frame Deph Video Super-Resoluion Kassem Al Ismaeil 1, Djamila Aouada 1, Thomas Solignac 2, Bruno Mirbach 2, Björn Oersen 1 1 Inerdisciplinary Cenre for Securiy, Reliabiliy, and

More information

Image Based Computer-Aided Manufacturing Technology

Image Based Computer-Aided Manufacturing Technology Sensors & Transducers 03 by IFSA hp://www.sensorsporal.com Image Based Compuer-Aided Manufacuring Technology Zhanqi HU Xiaoqin ZHANG Jinze LI Wei LI College of Mechanical Engineering Yanshan Universiy

More information

Detection Tracking and Recognition of Human Poses for a Real Time Spatial Game

Detection Tracking and Recognition of Human Poses for a Real Time Spatial Game Deecion Tracking and Recogniion of Human Poses for a Real Time Spaial Game Feifei Huo, Emile A. Hendriks, A.H.J. Oomes Delf Universiy of Technology The Neherlands f.huo@udelf.nl Pascal van Beek, Remco

More information

Sam knows that his MP3 player has 40% of its battery life left and that the battery charges by an additional 12 percentage points every 15 minutes.

Sam knows that his MP3 player has 40% of its battery life left and that the battery charges by an additional 12 percentage points every 15 minutes. 8.F Baery Charging Task Sam wans o ake his MP3 player and his video game player on a car rip. An hour before hey plan o leave, he realized ha he forgo o charge he baeries las nigh. A ha poin, he plugged

More information

Rao-Blackwellized Particle Filtering for Probing-Based 6-DOF Localization in Robotic Assembly

Rao-Blackwellized Particle Filtering for Probing-Based 6-DOF Localization in Robotic Assembly MITSUBISHI ELECTRIC RESEARCH LABORATORIES hp://www.merl.com Rao-Blackwellized Paricle Filering for Probing-Based 6-DOF Localizaion in Roboic Assembly Yuichi Taguchi, Tim Marks, Haruhisa Okuda TR1-8 June

More information

Tracking Appearances with Occlusions

Tracking Appearances with Occlusions Tracking ppearances wih Occlusions Ying Wu, Ting Yu, Gang Hua Deparmen of Elecrical & Compuer Engineering Norhwesern Universiy 2145 Sheridan oad, Evanson, IL 60208 {yingwu,ingyu,ganghua}@ece.nwu.edu bsrac

More information

Detection and segmentation of moving objects in highly dynamic scenes

Detection and segmentation of moving objects in highly dynamic scenes Deecion and segmenaion of moving objecs in highly dynamic scenes Aurélie Bugeau Parick Pérez INRIA, Cenre Rennes - Breagne Alanique Universié de Rennes, Campus de Beaulieu, 35 042 Rennes Cedex, France

More information

FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS

FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS Mohammed A. Aseeri and M. I. Sobhy Deparmen of Elecronics, The Universiy of Ken a Canerbury Canerbury, Ken, CT2

More information

Image Content Representation

Image Content Representation Image Conen Represenaion Represenaion for curves and shapes regions relaionships beween regions E.G.M. Perakis Image Represenaion & Recogniion 1 Reliable Represenaion Uniqueness: mus uniquely specify an

More information

Real-Time Avatar Animation Steered by Live Body Motion

Real-Time Avatar Animation Steered by Live Body Motion Real-Time Avaar Animaion Seered by Live Body Moion Oliver Schreer, Ralf Tanger, Peer Eiser, Peer Kauff, Bernhard Kaspar, and Roman Engler 3 Fraunhofer Insiue for Telecommunicaions/Heinrich-Herz-Insiu,

More information

MODEL BASED TECHNIQUE FOR VEHICLE TRACKING IN TRAFFIC VIDEO USING SPATIAL LOCAL FEATURES

MODEL BASED TECHNIQUE FOR VEHICLE TRACKING IN TRAFFIC VIDEO USING SPATIAL LOCAL FEATURES MODEL BASED TECHNIQUE FOR VEHICLE TRACKING IN TRAFFIC VIDEO USING SPATIAL LOCAL FEATURES Arun Kumar H. D. 1 and Prabhakar C. J. 2 1 Deparmen of Compuer Science, Kuvempu Universiy, Shimoga, India ABSTRACT

More information

Upper Body Tracking for Human-Machine Interaction with a Moving Camera

Upper Body Tracking for Human-Machine Interaction with a Moving Camera The 2009 IEEE/RSJ Inernaional Conference on Inelligen Robos and Sysems Ocober -5, 2009 S. Louis, USA Upper Body Tracking for Human-Machine Ineracion wih a Moving Camera Yi-Ru Chen, Cheng-Ming Huang, and

More information

Gender Classification of Faces Using Adaboost*

Gender Classification of Faces Using Adaboost* Gender Classificaion of Faces Using Adaboos* Rodrigo Verschae 1,2,3, Javier Ruiz-del-Solar 1,2, and Mauricio Correa 1,2 1 Deparmen of Elecrical Engineering, Universidad de Chile 2 Cener for Web Research,

More information

Robust Multi-view Face Detection Using Error Correcting Output Codes

Robust Multi-view Face Detection Using Error Correcting Output Codes Robus Muli-view Face Deecion Using Error Correcing Oupu Codes Hongming Zhang,2, Wen GaoP P, Xilin Chen 2, Shiguang Shan 2, and Debin Zhao Deparmen of Compuer Science and Engineering, Harbin Insiue of Technolog

More information

Nonparametric CUSUM Charts for Process Variability

Nonparametric CUSUM Charts for Process Variability Journal of Academia and Indusrial Research (JAIR) Volume 3, Issue June 4 53 REEARCH ARTICLE IN: 78-53 Nonparameric CUUM Chars for Process Variabiliy D.M. Zombade and V.B. Ghue * Dep. of aisics, Walchand

More information

Video Content Description Using Fuzzy Spatio-Temporal Relations

Video Content Description Using Fuzzy Spatio-Temporal Relations Proceedings of he 4s Hawaii Inernaional Conference on Sysem Sciences - 008 Video Conen Descripion Using Fuzzy Spaio-Temporal Relaions rchana M. Rajurkar *, R.C. Joshi and Sananu Chaudhary 3 Dep of Compuer

More information

Robust 3D Visual Tracking Using Particle Filtering on the SE(3) Group

Robust 3D Visual Tracking Using Particle Filtering on the SE(3) Group Robus 3D Visual Tracking Using Paricle Filering on he SE(3) Group Changhyun Choi and Henrik I. Chrisensen Roboics & Inelligen Machines, College of Compuing Georgia Insiue of Technology Alana, GA 3332,

More information

Viewpoint Invariant 3D Landmark Model Inference from Monocular 2D Images Using Higher-Order Priors

Viewpoint Invariant 3D Landmark Model Inference from Monocular 2D Images Using Higher-Order Priors Viewpoin Invarian 3D Landmark Model Inference from Monocular 2D Images Using Higher-Order Priors Chaohui Wang 1,2, Yun Zeng 3, Loic Simon 1, Ioannis Kakadiaris 4, Dimiris Samaras 3, Nikos Paragios 1,2

More information

Efficient Region Tracking With Parametric Models of Geometry and Illumination

Efficient Region Tracking With Parametric Models of Geometry and Illumination EEE TRANSACTONS ON PATTERN ANALYSS AND MACHNE NTELLGENCE, VOL. 2, NO. 1, OCTOBER 1998 1 Efficien Region Tracking Wih Parameric Models of Geomery and lluminaion Gregory D. Hager, Member, EEE, and Peer N.

More information

Curves & Surfaces. Last Time? Today. Readings for Today (pick one) Limitations of Polygonal Meshes. Today. Adjacency Data Structures

Curves & Surfaces. Last Time? Today. Readings for Today (pick one) Limitations of Polygonal Meshes. Today. Adjacency Data Structures Las Time? Adjacency Daa Srucures Geomeric & opologic informaion Dynamic allocaion Efficiency of access Curves & Surfaces Mesh Simplificaion edge collapse/verex spli geomorphs progressive ransmission view-dependen

More information

LAMP: 3D Layered, Adaptive-resolution and Multiperspective Panorama - a New Scene Representation

LAMP: 3D Layered, Adaptive-resolution and Multiperspective Panorama - a New Scene Representation Submission o Special Issue of CVIU on Model-based and Image-based 3D Scene Represenaion for Ineracive Visualizaion LAMP: 3D Layered, Adapive-resoluion and Muliperspecive Panorama - a New Scene Represenaion

More information

Low-Cost WLAN based. Dr. Christian Hoene. Computer Science Department, University of Tübingen, Germany

Low-Cost WLAN based. Dr. Christian Hoene. Computer Science Department, University of Tübingen, Germany Low-Cos WLAN based Time-of-fligh fligh Trilaeraion Precision Indoor Personnel Locaion and Tracking for Emergency Responders Third Annual Technology Workshop, Augus 5, 2008 Worceser Polyechnic Insiue, Worceser,

More information

In fmri a Dual Echo Time EPI Pulse Sequence Can Induce Sources of Error in Dynamic Magnetic Field Maps

In fmri a Dual Echo Time EPI Pulse Sequence Can Induce Sources of Error in Dynamic Magnetic Field Maps In fmri a Dual Echo Time EPI Pulse Sequence Can Induce Sources of Error in Dynamic Magneic Field Maps A. D. Hahn 1, A. S. Nencka 1 and D. B. Rowe 2,1 1 Medical College of Wisconsin, Milwaukee, WI, Unied

More information

4. Minimax and planning problems

4. Minimax and planning problems CS/ECE/ISyE 524 Inroducion o Opimizaion Spring 2017 18 4. Minima and planning problems ˆ Opimizing piecewise linear funcions ˆ Minima problems ˆ Eample: Chebyshev cener ˆ Muli-period planning problems

More information

A time-space consistency solution for hardware-in-the-loop simulation system

A time-space consistency solution for hardware-in-the-loop simulation system Inernaional Conference on Advanced Elecronic Science and Technology (AEST 206) A ime-space consisency soluion for hardware-in-he-loop simulaion sysem Zexin Jiang a Elecric Power Research Insiue of Guangdong

More information

Gauss-Jordan Algorithm

Gauss-Jordan Algorithm Gauss-Jordan Algorihm The Gauss-Jordan algorihm is a sep by sep procedure for solving a sysem of linear equaions which may conain any number of variables and any number of equaions. The algorihm is carried

More information

Spline Curves. Color Interpolation. Normal Interpolation. Last Time? Today. glshademodel (GL_SMOOTH); Adjacency Data Structures. Mesh Simplification

Spline Curves. Color Interpolation. Normal Interpolation. Last Time? Today. glshademodel (GL_SMOOTH); Adjacency Data Structures. Mesh Simplification Las Time? Adjacency Daa Srucures Spline Curves Geomeric & opologic informaion Dynamic allocaion Efficiency of access Mesh Simplificaion edge collapse/verex spli geomorphs progressive ransmission view-dependen

More information

Robust Segmentation and Tracking of Colored Objects in Video

Robust Segmentation and Tracking of Colored Objects in Video IEEE TRANSACTIONS ON CSVT, VOL. 4, NO. 6, 2004 Robus Segmenaion and Tracking of Colored Objecs in Video Theo Gevers, member, IEEE Absrac Segmening and racking of objecs in video is of grea imporance for

More information

Moving Object Detection Using MRF Model and Entropy based Adaptive Thresholding

Moving Object Detection Using MRF Model and Entropy based Adaptive Thresholding Moving Objec Deecion Using MRF Model and Enropy based Adapive Thresholding Badri Narayan Subudhi, Pradipa Kumar Nanda and Ashish Ghosh Machine Inelligence Uni, Indian Saisical Insiue, Kolkaa, 700108, India,

More information

An Adaptive Spatial Depth Filter for 3D Rendering IP

An Adaptive Spatial Depth Filter for 3D Rendering IP JOURNAL OF SEMICONDUCTOR TECHNOLOGY AND SCIENCE, VOL.3, NO. 4, DECEMBER, 23 175 An Adapive Spaial Deph Filer for 3D Rendering IP Chang-Hyo Yu and Lee-Sup Kim Absrac In his paper, we presen a new mehod

More information

We are IntechOpen, the world s leading publisher of Open Access books Built by scientists, for scientists. International authors and editors

We are IntechOpen, the world s leading publisher of Open Access books Built by scientists, for scientists. International authors and editors We are InechOpen, he world s leading publisher of Open Access books Buil by scieniss, for scieniss 4,000 116,000 120M Open access books available Inernaional auhors and ediors Downloads Our auhors are

More information

Assignment 2. Due Monday Feb. 12, 10:00pm.

Assignment 2. Due Monday Feb. 12, 10:00pm. Faculy of rs and Science Universiy of Torono CSC 358 - Inroducion o Compuer Neworks, Winer 218, LEC11 ssignmen 2 Due Monday Feb. 12, 1:pm. 1 Quesion 1 (2 Poins): Go-ack n RQ In his quesion, we review how

More information

Large-scale 3D Outdoor Mapping and On-line Localization using 3D-2D Matching

Large-scale 3D Outdoor Mapping and On-line Localization using 3D-2D Matching Large-scale 3D Oudoor Mapping and On-line Localizaion using 3D-D Maching Takahiro Sakai, Kenji Koide, Jun Miura, and Shuji Oishi Absrac Map-based oudoor navigaion is an acive research area in mobile robos

More information

MATH Differential Equations September 15, 2008 Project 1, Fall 2008 Due: September 24, 2008

MATH Differential Equations September 15, 2008 Project 1, Fall 2008 Due: September 24, 2008 MATH 5 - Differenial Equaions Sepember 15, 8 Projec 1, Fall 8 Due: Sepember 4, 8 Lab 1.3 - Logisics Populaion Models wih Harvesing For his projec we consider lab 1.3 of Differenial Equaions pages 146 o

More information

Proceeding of the 6 th International Symposium on Artificial Intelligence and Robotics & Automation in Space: i-sairas 2001, Canadian Space Agency,

Proceeding of the 6 th International Symposium on Artificial Intelligence and Robotics & Automation in Space: i-sairas 2001, Canadian Space Agency, Proceeding of he 6 h Inernaional Symposium on Arificial Inelligence and Roboics & Auomaion in Space: i-sairas 00, Canadian Space Agency, S-Huber, Quebec, Canada, June 8-, 00. Muli-resoluion Mapping Using

More information

A High-Speed Adaptive Multi-Module Structured Light Scanner

A High-Speed Adaptive Multi-Module Structured Light Scanner A High-Speed Adapive Muli-Module Srucured Ligh Scanner Andreas Griesser 1 Luc Van Gool 1,2 1 Swiss Fed.Ins.of Techn.(ETH) 2 Kaholieke Univ. Leuven D-ITET/Compuer Vision Lab ESAT/VISICS Zürich, Swizerland

More information

Weighted Voting in 3D Random Forest Segmentation

Weighted Voting in 3D Random Forest Segmentation Weighed Voing in 3D Random Fores Segmenaion M. Yaqub,, P. Mahon 3, M. K. Javaid, C. Cooper, J. A. Noble NDORMS, Universiy of Oxford, IBME, Deparmen of Engineering Science, Universiy of Oxford, 3 MRC Epidemiology

More information

A Hierarchical Object Recognition System Based on Multi-scale Principal Curvature Regions

A Hierarchical Object Recognition System Based on Multi-scale Principal Curvature Regions A Hierarchical Objec Recogniion Sysem Based on Muli-scale Principal Curvaure Regions Wei Zhang, Hongli Deng, Thomas G Dieerich and Eric N Morensen School of Elecrical Engineering and Compuer Science Oregon

More information

Michiel Helder and Marielle C.T.A Geurts. Hoofdkantoor PTT Post / Dutch Postal Services Headquarters

Michiel Helder and Marielle C.T.A Geurts. Hoofdkantoor PTT Post / Dutch Postal Services Headquarters SHORT TERM PREDICTIONS A MONITORING SYSTEM by Michiel Helder and Marielle C.T.A Geurs Hoofdkanoor PTT Pos / Duch Posal Services Headquarers Keywords macro ime series shor erm predicions ARIMA-models faciliy

More information

Motion Estimation of a Moving Range Sensor by Image Sequences and Distorted Range Data

Motion Estimation of a Moving Range Sensor by Image Sequences and Distorted Range Data Moion Esimaion of a Moving Range Sensor by Image Sequences and Disored Range Daa Asuhiko Banno, Kazuhide Hasegawa and Kasushi Ikeuchi Insiue of Indusrial Science Universiy of Tokyo 4-6-1 Komaba, Meguro-ku,

More information

A new algorithm for small object tracking based on super-resolution technique

A new algorithm for small object tracking based on super-resolution technique A new algorihm for small objec racking based on super-resoluion echnique Yabunayya Habibi, Dwi Rana Sulisyaningrum, and Budi Seiyono Ciaion: AIP Conference Proceedings 1867, 020024 (2017); doi: 10.1063/1.4994427

More information

X-Splines : A Spline Model Designed for the End-User

X-Splines : A Spline Model Designed for the End-User X-Splines : A Spline Model Designed for he End-User Carole Blanc Chrisophe Schlic LaBRI 1 cours de la libéraion, 40 alence (France) [blancjschlic]@labri.u-bordeaux.fr Absrac his paper presens a new model

More information

Audio Engineering Society. Convention Paper. Presented at the 119th Convention 2005 October 7 10 New York, New York USA

Audio Engineering Society. Convention Paper. Presented at the 119th Convention 2005 October 7 10 New York, New York USA Audio Engineering Sociey Convenion Paper Presened a he 119h Convenion 2005 Ocober 7 10 New Yor, New Yor USA This convenion paper has been reproduced from he auhor's advance manuscrip, wihou ediing, correcions,

More information

STRING DESCRIPTIONS OF DATA FOR DISPLAY*

STRING DESCRIPTIONS OF DATA FOR DISPLAY* SLAC-PUB-383 January 1968 STRING DESCRIPTIONS OF DATA FOR DISPLAY* J. E. George and W. F. Miller Compuer Science Deparmen and Sanford Linear Acceleraor Cener Sanford Universiy Sanford, California Absrac

More information

Mobile Robots Mapping

Mobile Robots Mapping Mobile Robos Mapping 1 Roboics is Easy conrol behavior percepion modelling domain model environmen model informaion exracion raw daa planning ask cogniion reasoning pah planning navigaion pah execuion

More information

Recovering Joint and Individual Components in Facial Data

Recovering Joint and Individual Components in Facial Data JOURNAL OF L A E X CLASS FILES, VOL. 14, NO. 8, AUGUS 2015 1 Recovering Join and Individual Componens in Facial Daa Chrisos Sagonas, Evangelos Ververas, Yannis Panagakis, and Sefanos Zafeiriou, Member,

More information

Definition and examples of time series

Definition and examples of time series Definiion and examples of ime series A ime series is a sequence of daa poins being recorded a specific imes. Formally, le,,p be a probabiliy space, and T an index se. A real valued sochasic process is

More information

Streamline Pathline Eulerian Lagrangian

Streamline Pathline Eulerian Lagrangian Sreamline Pahline Eulerian Lagrangian Sagnaion Poin Flow V V V = + = + = + o V xi y j a V V xi y j o Pahline and Sreakline Insananeous Sreamlines Pahlines Sreaklines Maerial Derivaive Acceleraion

More information

Last Time: Curves & Surfaces. Today. Questions? Limitations of Polygonal Meshes. Can We Disguise the Facets?

Last Time: Curves & Surfaces. Today. Questions? Limitations of Polygonal Meshes. Can We Disguise the Facets? Las Time: Curves & Surfaces Expeced value and variance Mone-Carlo in graphics Imporance sampling Sraified sampling Pah Tracing Irradiance Cache Phoon Mapping Quesions? Today Moivaion Limiaions of Polygonal

More information

Track and Cut: simultaneous tracking and segmentation of multiple objects with graph cuts

Track and Cut: simultaneous tracking and segmentation of multiple objects with graph cuts INSTITUT NATIONAL DE RECHERCHE EN INFORMATIQUE ET EN AUTOMATIQUE Track and Cu: simulaneous racking and segmenaion of muliple objecs wih graph cus Aurélie Bugeau Parick Pérez N 6337 Ocober 2007 Thèmes COM

More information

Computer representations of piecewise

Computer representations of piecewise Edior: Gabriel Taubin Inroducion o Geomeric Processing hrough Opimizaion Gabriel Taubin Brown Universiy Compuer represenaions o piecewise smooh suraces have become vial echnologies in areas ranging rom

More information

Motion Level-of-Detail: A Simplification Method on Crowd Scene

Motion Level-of-Detail: A Simplification Method on Crowd Scene Moion Level-of-Deail: A Simplificaion Mehod on Crowd Scene Absrac Junghyun Ahn VR lab, EECS, KAIST ChocChoggi@vr.kais.ac.kr hp://vr.kais.ac.kr/~zhaoyue Recen echnological improvemen in characer animaion

More information

An Improved Square-Root Nyquist Shaping Filter

An Improved Square-Root Nyquist Shaping Filter An Improved Square-Roo Nyquis Shaping Filer fred harris San Diego Sae Universiy fred.harris@sdsu.edu Sridhar Seshagiri San Diego Sae Universiy Seshigar.@engineering.sdsu.edu Chris Dick Xilinx Corp. chris.dick@xilinx.com

More information

RGBD Data Based Pose Estimation: Why Sensor Fusion?

RGBD Data Based Pose Estimation: Why Sensor Fusion? 18h Inernaional Conference on Informaion Fusion Washingon, DC - July 6-9, 2015 RGBD Daa Based Pose Esimaion: Why Sensor Fusion? O. Serdar Gedik Deparmen of Compuer Engineering, Yildirim Beyazi Universiy,

More information

Tracking Deforming Objects Using Particle Filtering for Geometric Active Contours

Tracking Deforming Objects Using Particle Filtering for Geometric Active Contours 1470 IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 29, NO. 8, AUGUST 2007 Tracking Deforming Objecs Using Paricle Filering for Geomeric Acive Conours Yogesh Rahi, Member, IEEE, NamraaVaswani,

More information

An efficient approach to improve throughput for TCP vegas in ad hoc network

An efficient approach to improve throughput for TCP vegas in ad hoc network Inernaional Research Journal of Engineering and Technology (IRJET) e-issn: 395-0056 Volume: 0 Issue: 03 June-05 www.irje.ne p-issn: 395-007 An efficien approach o improve hroughpu for TCP vegas in ad hoc

More information

IntentSearch:Capturing User Intention for One-Click Internet Image Search

IntentSearch:Capturing User Intention for One-Click Internet Image Search JOURNAL OF L A T E X CLASS FILES, VOL. 6, NO. 1, JANUARY 2010 1 InenSearch:Capuring User Inenion for One-Click Inerne Image Search Xiaoou Tang, Fellow, IEEE, Ke Liu, Jingyu Cui, Suden Member, IEEE, Fang

More information

Collision-Free and Curvature-Continuous Path Smoothing in Cluttered Environments

Collision-Free and Curvature-Continuous Path Smoothing in Cluttered Environments Collision-Free and Curvaure-Coninuous Pah Smoohing in Cluered Environmens Jia Pan 1 and Liangjun Zhang and Dinesh Manocha 3 1 panj@cs.unc.edu, 3 dm@cs.unc.edu, Dep. of Compuer Science, Universiy of Norh

More information

COSC 3213: Computer Networks I Chapter 6 Handout # 7

COSC 3213: Computer Networks I Chapter 6 Handout # 7 COSC 3213: Compuer Neworks I Chaper 6 Handou # 7 Insrucor: Dr. Marvin Mandelbaum Deparmen of Compuer Science York Universiy F05 Secion A Medium Access Conrol (MAC) Topics: 1. Muliple Access Communicaions:

More information

Graffiti Detection Using Two Views

Graffiti Detection Using Two Views Graffii Deecion Using wo Views Luigi Di Sefano Federico ombari Alessandro Lanza luigi.disefano@unibo.i federico.ombari@unibo.i alanza@arces.unibo.i Sefano Maoccia sefano.maoccia@unibo.i Sefano Moni sefano.moni3@sudio.unibo.i

More information

M(t)/M/1 Queueing System with Sinusoidal Arrival Rate

M(t)/M/1 Queueing System with Sinusoidal Arrival Rate 20 TUTA/IOE/PCU Journal of he Insiue of Engineering, 205, (): 20-27 TUTA/IOE/PCU Prined in Nepal M()/M/ Queueing Sysem wih Sinusoidal Arrival Rae A.P. Pan, R.P. Ghimire 2 Deparmen of Mahemaics, Tri-Chandra

More information

Effects needed for Realism. Ray Tracing. Ray Tracing: History. Outline. Foundations of Computer Graphics (Fall 2012)

Effects needed for Realism. Ray Tracing. Ray Tracing: History. Outline. Foundations of Computer Graphics (Fall 2012) Foundaions of ompuer Graphics (Fall 2012) S 184, Lecure 16: Ray Tracing hp://ins.eecs.berkeley.edu/~cs184 Effecs needed for Realism (Sof) Shadows Reflecions (Mirrors and Glossy) Transparency (Waer, Glass)

More information