Robust Visual Tracking for Multiple Targets

Size: px
Start display at page:

Download "Robust Visual Tracking for Multiple Targets"

Transcription

1 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, Absrac. We address he problem of robus muli-arge racking wihin he applicaion of hockey player racking. The paricle filer echnique is adoped and modified o fi ino he muli-arge racking framework. A recificaion echnique is employed o find he correspondence beween he video frame coordinaes and he sandard hockey rink coordinaes so ha he sysem can compensae for camera moion and improve he dynamics of he players. A global neares neighbor daa associaion algorihm is inroduced o assign boosing deecions o he exising racks for he proposal disribuion in paricle filers. The meanshif algorihm is embedded ino he paricle filer framework o sabilize he rajecories of he arges for robus racking during muual occlusion. Experimenal resuls show ha our sysem is able o auomaically and robusly rack a variable number of arges and correcly mainain heir ideniies regardless of background cluer, camera moion and frequen muual occlusion beween arges. 1 Inroducion Tracking muliple arges, alhough has is roo in conrol heory, has been of broad ineres in many compuer vision applicaions for decades as well. A visualbased muli-arge racking sysem should be able o rack a variable number of objecs in a dynamic scene and mainain he correc ideniies of he arges regardless of occlusions and any oher visual perurbaions. As i is a very complicaed and challenging problem, exensive research work has been done. In his work, we address he problem of robus muli-arge racking wihin he applicaion of hockey player racking. Paricle filering was firs inroduced o visual racking by Isard and Blake in [1]. Pérez e al. [2, 3] exended he paricle filer framework o rack muliple arges. Okuma e al. [4] furher exended i [3] by incorporaing a boosing deecor [5] ino he paricle filer for auomaic iniializaion of a variable number of arges. However, as heir sysem did no have explici mechanisms o model muual occlusions beween arges, i loses he ideniies of he arges afer occlusions. On he oher hand, various approaches have been aken o solve he occlusion problem in racking. Kang e al. [6] ried o resolve he ambiguiy of he locaions of he arges by regisering video frames from muliple cameras. Zhao e al. [7] also recified video frames o he predefined ground plane and model he arges in he 3D space wih a body shape model. A saic camera was used and background subracion was applied as well in heir work. Explici A. Leonardis, H. Bischof, and A. Pinz (Eds.): ECCV 2006, Par IV, LNCS 3954, pp , c Springer-Verlag Berlin Heidelberg 2006

2 108 Y. Cai, N. de Freias, and J.J. Lile arge shape modelling can help resolving he likelihood modelling and daa associaion problems during occlusions. The approach is ofen used wihin saic scenes [8, 9, 10]. However, in our applicaion, camera moion makes i difficul o separae arge moion or perform background subracion. Players wih drasic pose changes are difficul o be capured by any explici shape models. In order o build a racking sysem ha can correcly rack muliple arges regardless of camera moion and muual occlusion, we propose four improvemens on he previous sysems. Firsly, a recificaion echnique is employed o compensae for camera moions. Secondly, a second order auoregression model is adoped as he dynamics model. Thirdly, a global neares neighbor daa associaion echnique is used o correcly associae boosing deecions wih he exising racks. Finally, he mean-shif algorihm is embedded ino he paricle filer framework o sabilize he rajecories of he arges for reliable moion predicion. Alhough similar work [11] has been done on combining mean-shif wih paricle filering, our work is he firs one ha describes in deail he heoreical formulaion of embedding mean-shif seamlessly ino he paricle filer framework for muli-arge racking. Consequenly, alhough our sysem performs comparably o he sysem in [4], i significanly improves upon ha sysem when occlusions happen, which is he main focus of his work. 2 Filering Paricle filering has been a successful numerical approximaion echnique for Bayesian sequenial esimaion wih non-linear, non-gaussian models. In our applicaion, he fas moion of hockey players and he color model we adop [12, 13] is highly non-linear and non-gaussian. Therefore, paricle filering is he ideal model o be he basic skeleon of our racking sysem. The basic Bayesian filering is a recursive process in which each ieraion consiss of a predicion sep and a filering sep. predicion sep: p(x y 0: 1 )= p(x x 1 )p(x 1 y 0: 1 )dx 1 filering sep: p(x y 0: )= p(y x)p(x y0: 1) p(y x )p(x y 0: 1)dx (1) where he process is iniialized by he prior disribuion p(x 0 y 0 ) = p(x 0 ), p(x x 1 ) is he arge dynamics model, and p(y x ) is he likelihood model. Paricle filering uses a se of weighed samples {x (i),w (i) } Ns i=1 o approximae he poserior disribuion in he filering. The sample se is propagaed by sampling from a designed proposal disribuion q(x x 1,y 0: ), which is called imporance sampling. The imporance weighs of he paricles are updaed in each ieraion as follows w (i) p(y x (i) )p(x (i) x (i) 1 ) q(x (i) x (i) 1,y 0:) Ns w (i) 1, i=1 w (i) =1. (2) Resampling of he paricles is necessary from ime o ime in each ieraion o avoid degeneracy of he imporance weighs.

3 Robus Visual Tracking for Muliple Targes 109 One of he criical issues in keeping paricle filering effecive is he design of he proposal disribuion. The proposal disribuion should be able o shif he paricles o he regions wih high likelihood if here is a big gap beween he mode of he prior disribuion and he mode of likelihood disribuion. The boosed paricle filer (BPF) [4] used a mixure of Gaussians model ha combines boh he dynamics prior and he Adaboos deecions [5] q B(x x 1,y 0: )=αq ada (x y )+(1 α)p(x x 1 ), (3) where α is he parameer ha is dynamically updaed according o he overlap beween he Gaussian disribuion of boosing deecion and he dynamics prior. The issue of daa associaion arises here. Deails abou how o correcly assign boosing deecions o he exising racks will be discussed laer. In addiion, he original BPF work by Okuma e al. [4] is based on he mixure of paricle filer srucure (MPF) [3], which has a fixed number of paricles for all he arges. As a resul, new arges have o seal paricles from exising racks and reduce he accuracy of he approximaion. The merge and spli of paricle clouds in he MPF srucure also cause he loss of he correc ideniies of he arges during occlusions. Therefore, we adop he boosed paricle filer as he basic filering framework in our applicaion. However, insead of using he MPF srucure, we use an independen paricle se for each arge o avoid he wo inheren disadvanages of MPF. 3 Targe Dynamics Modelling In visual racking sysems, accurae modelling of he arge dynamics can improve he predicion of he locaions of he arges while visual suppor is insufficien due o occlusion. However, because of he camera moion in our applicaion, he image coordinae sysem changes over ime wih respec o he hockey rink coordinaes. Therefore, arge moion modelling and predicion in he image coordinaes are difficul. We adop he approach by Okuma e al. [14] o map he locaions of he arges from he image coordinaes o he sandard hockey rink coordinae sysem which is consisen over ime. Therefore, according o he physical law of ineria, he moions of he players in hockey games are beer prediced wih a consan velociy auoregressive model. 3.1 Recificaion Homography is defined by Harley and Zisserman in [15] as an inverible mapping h beween wo planes. Images recorded by cameras are 2D projecions of he real world. For any plane in he world, is images from a camera, which can pan, il, zoom or even move, are exacly modelled by a homography as long as here is no noiceable lens disorion. As he hockey players are always moving in he plane formed by he hockey rink, heir locaions on he rink are in he same plane boh in he real world and he image space. Therefore, i is possible o projec heir locaions beween he wo planes.

4 110 Y. Cai, N. de Freias, and J.J. Lile Fig. 1. This shows a projeced video frame blended wih he sandard hockey rink The work by Okuma e al. [14] is able o auomaically compue he homography beween video frames and he hockey rink. Figure 1 shows how he video frames are mapped o he sandard rink wih he homography. Wih his homography, he hidden saes of he arges are represened in he rink coordinaes and paricle filering is performed in he rink coordinaes as well. Hidden saes will be mapped o he image coordinaes when evaluaing he likelihood of he observaion. 3.2 Auoregressive Dynamics Model An auoregressive process is a ime series modelling sraegy which akes ino accoun he hisorical daa o predic he curren sae value. In his model, he curren sae x only depends on he previous saes wih a deerminisic mapping funcion and a sochasic disurbance. As he paricle filering process is performed in he sandard rink coordinaes, he moions of he players on he rink are separaed from he camera moion. Thus, he modelling is much easier. In hockey games, because of he effec of ineria, a consan velociy model is suiable o model he moion of he players. I is bes described by he following second order auoregressive model x = Ax 1 + Bx 2 + CN (0,Σ) (4) where {A, B, C} are he auoregression coefficiens, N (0,Σ) is a Gaussian noise wih zero mean and sandard deviaion of 1. 4 Daa Associaion In a sandard Bayesian filering framework, daa associaion is performed o pair he observaions and racks for he evaluaion of he likelihood funcion p(y m x n ). Wih proper esimaion of segmenaion and shape of he arges [10], he observaion can be assigned o racks in a globally opimal way. However, as we do no have an explici shape model for he arges, he paricle filer framework in our applicaion handles his level of daa associaion locally in

5 Robus Visual Tracking for Muliple Targes 111 an implici way. Because he boosing deecions are used o improve he proposal disribuion in paricle filers as in shown in Equaion 3, we perform daa associaion a his level o assign boosing deecions o he exising racks. 4.1 Linear Opimizaion The assignmen problem can be bes represened by an assignmen marix shown in Table 1. Each enry in he able is he cos or gain of pairing he corresponding rack and observaion. In our applicaion, he values of all he enries in he assignmen marix are defined o be he disance beween he observaions and he racks in he rink coordinaes. Assignmens ha are forbidden by gaing are denoed by in he corresponding enries. Observaions ha are forbidden by he gaing o be associaed o any rack are considered as a new rack in our applicaion. Table 1. Example of he assignmen marix for he assignmen problem Observaions Tracks O1 O2 O3 O4 T1 a 11 a 12 T2 a 21 a 24 T3 a 31 a 34 Such assignmen problems sem from economic heory and aucion heory as well. The objecive is o minimize he cos or maximize he gain subjec o a se of consrains. Given he assignmen marix shown in Table 1, he objecive is o find a se X = {x ij }, which are binary indicaors, ha maximizes or minimizes he objecive funcion C = n m i=1 j=1 a ijx ij subjec o some linear consrains. Linear programming was iniially used o solve his problem. Laer on, i was found ha he aucion algorihm [16] is he mos efficien mehod so far o reach he opimal soluion or sub-opimal one wihou any pracical difference. The exended aucion algorihm [17] is able o solve he recangular marix problems wih he consrain ha one observaion can only be assigned o one arge while a arge can have a leas one observaions. However, in our applicaion, i is very likely ha some racks do no have any observaion due o he mis-deecion of he boosing deecor. Therefore, even if here are some observaions wihin he gae of ha rack, i is sill possible ha none of he observaions belongs o he rack. Hence, he consrains are formalized as n i=1 x ij =1, j m j=1 x ij 0, i (5) and he soluion is x i j = { 1ifi =arg i min a ij 0 oherwise j (6)

6 112 Y. Cai, N. de Freias, and J.J. Lile 5 Mean-Shif Embedded Paricle Filer The moivaion of embedding he mean-shif algorihm ino he paricle filer framework of our racking sysem is o sabilize he racking resul. I is imporan for he dynamics model because sabilizing rajecories improves he accuracy of he compued velociy of arges, which is criical for improving he predicion of he locaion of he arges. I is also imporan for he likelihood model because accurae predicion leads sampling o more promising areas so ha he influence from background cluer and muual occlusion will be reduced. 5.1 Color Model We adoped he color model in [13, 4] in our applicaion because i is successful in racking non-rigid objecs wih parial occlusion. The model is originally inroduced by Comaniciu e al. [18] for he mean-shif based objec racking. The observaion of he arge is represened by an N-bin color hisogram exraced from he region R(x ) cenered a he locaion x. I is denoed as Q(x )={q(n; x )} n=1,...,n, where q(n; x )=C k R(x ) δ[b(k) n] (7) where δ is he Kronecker dela funcion, C is a normalizaion consan, k is any pixel wihin he region R(x ). By normalizing he color hisogram, Q(x ) becomes a discree probabilisic disribuion. The similariy beween he curren observaion Q(x ) and he reference model Q, which is consruced a he iniializaion sep, is evaluaed based on he Bhaacharyya coefficien d(x, x 0 )= 1 ρ[q(x ),Q ],ρ[q(x ),Q ]= N q(n; x )q (n; x 0 ) (8) In order o encode he spaial informaion of he observaion, a muli-par color model [13, 4] is employed, which splis he arges verically ino wo pars. The color hisogram of he wo pars are consruced separaely and concaenaed in parallel as a new hisogram. The likelihood is hen evaluaed as n=1 p(y x ) e λd2 (x,x 0). (9) 5.2 Mean-Shif Mean-shif is a nonparameric saisical mehod ha seeks he mode of a densiy disribuion in an ieraive procedure. I was firs generalized and analyzed by Cheng in [19] and laer developed by Comaniciu e al. in [20]. The objecive of he mean-shif algorihm is o ieraively shif he curren locaion x o a new locaion x according o he following relaion ( M x i=1 a a iw(a i )k i x 2) h = ( M i=1 w(a a i)k i x 2) (10) h

7 Robus Visual Tracking for Muliple Targes 113 where {a i } M i=1 are normalized poins wihin he region R(x) around he curren locaion x, w(a i ) is he weigh associaed o each pixel a i,andk(x) is a kernel profile of kernel K ha can be wrien in erms of a profile funcion k :[0, ) R such ha K(x) =k( x 2 ). According o [19], he kernel profile k(x) should be nonnegaive, nonincreasing, piecewise coninuous, and k(r)dr <. 0 The heory guaranees ha he mean-shif offse a locaion x is in he opposie direcion of he gradien direcion of he convoluion surface C(x) = M G(a i x)w(a i ) (11) i=1 where kernel G is called he shadow of kernel K and profile k(x) is proporional o he derivaive of profile g(x). In order o uilize mean-shif o analyze a discree densiy disribuion, i.e., he color hisogram, an isoropic kernel G wih a convex and monoonically decreasing kernel profile g(x) is superimposed ono he candidae region R(x ) o consruc such a convoluion surface. Therefore, he new color model can be rewrien as M h q(n; x )=C h i=1 g ( a i x h 2) δ[b(a i ) n] (12) where C h is also a normalizaion consan ha depends on h, andh is he bandwidh ha deermines he scale of he arge candidae. I should be noed ha in our applicaion, scale of he arges is separaed from he sae space of he arges and smoohly updaed, on per paricle basis, using he adapive scaling sraegy in [12]. The weigh in he mean-shif updae for he color feaure is shown below. N q w(a i )= (n; x 0 ) q(n; x) δ[b(a i) n]. (13) n=1 The Epanechnikov profile [12] is chosen o be he kernel profile of kernel G in our applicaion. Because i is linear, he kernel K becomes a consan and he kernel erm in Equaion 13 can be omied. 5.3 Mean-Shif Embedded Paricle Filer Applying he mean-shif algorihm direcly o he racking oupu only gives one deerminisic offse a each sep. I migh no be able o capure he rue locaion of he arges due o background cluer or muual occlusion beween arges in he image. Embedding i ino he paricle filer framework brings uncerainy o he deerminisic mehod so ha he saisical propery can improve he robusness of he algorihm. In our applicaion, he mean-shif operaion biases all he paricles righ afer he sampling from he mixure of Gaussians proposal disribuion and before he resampling sep in he paricle filer framework. Alhough similar work [11] has been done for racking, i was only for single arge and he proper way of updaing he paricle weighs afer he mean-shif bias was no addressed clearly.

8 114 Y. Cai, N. de Freias, and J.J. Lile However, embedding he mean-shif algorihm seamlessly ino he paricle filer framework wihou inroducing bias is non-rivial. Direcly biasing sampled paricles from he old proposal disribuion will change he overall poserior disribuion. This makes updaing he weigh of he paricles wihou bias exremely difficul. Alhough he mean-shif bias is a deerminisic mapping so ha i can be seen as a change of variable, i is no applicable in pracice. On one hand, because he mean-shif bias is a muliple o one mapping, i is no inverible. On he oher hand, because i is difficul o wrie he mean-shif bias in an analyical expression for differeniaion even in a piecewise manner, i is difficul o compue he Jacobian marix in he variable change. We ake an alernaive approach in our applicaion. Mean-shif biases he } i=1,...,n ha are propagaed by he old proposal disribuion o a new paricle se { x (i) } i=1,...,n. We denoe mean-shif searching wih funcion ϕ( ) such ha x = ϕ(ˆx ). Finally, a Gaussian disribuion is superimposed on he biased paricles o sample new paricles. Therefore, he mean-shif bias wih a superimposed Gaussian disribuion combined wih he old proposal disribuion can be considered as a new proposal disribuion q(x x 1, y ). For he new proposal disribuion, he weigh is updaed as follows: samples {ˆx (i) w (i) p(y x (i) )p( x (i) x (i) 1 ) q( x (i) x (i) 1, y ) w (i) 1 (14) where q( x (i) x (i) 1, y )=N ( x (i) x (i),σ). Here, Σ is a diagonal 2 2 marix and he value of he wo enries are chosen o be he same, which is 0.3, in our applicaion. Noe ha we use a sample x (i) insead of he biased paricle x (i). This ensures ha he sequenial imporance sampler remains unbiased and valid. The following pseudo-code depics he overall srucure of our racking sysem, which includes all he conribuions in our work. Iniializaion: =0 Map boosing deecions o he rink coordinaes o ge {x k,0 } k=1,...,m0. Creae paricle se {x (i) k,0, 1 N }N i=1 by sampling from p(x k,0). For =1,..., T, 1. Targes addiion and removal Remove arges wih large paricle se variance. Map boosing deecions from he video frame o he rink. Daa associaion Creae a paricle se for each new arge. Associae boosing deecions o he exising racks o consruc Gaussian mixure proposal disribuion q(x k, x k, 1, z k, ), where z k, is boosing deecion. 2. For all paricles in each rack Imporance sampling For all paricles in each rack, sample ˆx (i) k, q(x k, x (i) k, 1, z k,).

9 Robus Visual Tracking for Muliple Targes 115 Mean-shif biasing Bias he paricles as x (i) k, = ϕ(ˆx(i) k, ). Sample x (i) k, q(x k, x (i) k, ) Weigh updae Updae weighs w (i) k, according o Equaion 14 and normalize. 3. Deerminisic resampling For each rack, resample paricles o ge new sample se {x (i) k,, 1 N }N i=1. 4. Oupu For each rack, E(x k, )= N i=1 w(i) k, x(i) k,. 6 Experimenal Resuls Figure 2 shows he comparison beween he racking resuls of he sysem in [4] and our sysem. Subfigure (a) is he key frame in he same racking sequence ha shows he overall view of he racking resuls. Subfigures (b-e) and (f-i) are he close-up views of he recangular region labelled in (a). Each player has a unique color box assigned o i. The color of he same player may no necessarily he same across resuls of he wo sysems. According o he resuls, we can see from Subfigures (b-e) ha he rackers merge ogeher when hey ge close and a new rack is creaed when hey spli. Meanwhile, our sysem can mainain correc ideniies during occlusion. Subfigures (j-u) in Figure 2 shows he paricle represenaion of he racking resuls of our sysem. In he pseudo-code in Secion 5.3, he evoluion of paricle ses in each ieraion of propagaion can be divided ino hree seps: before he mean-shif bias, afer he bias, and afer he deerminisic resampling. The las hree rows in he figure compare he difference beween he paricle ses afer each sep. Generally, he mean-shif algorihm moves paricles from differen locaions around he arge o locaions in he neighborhood ha are mos similar o he reference model in he color space. Therefore, paricle ses appear more condensed afer he mean-shif bias. The difference beween Subfigure (p) and (q) in Figure 2 indicaes ha mean-shif migh move paricles o some oher arges because of he similariy beween he wo arges in he color space. However, such paricles will be assigned low weighs because of he regularizaion of he dynamics model. As a resul, hose paricles will have much fewer or no children afer he resampling sage. For he same reason, paricles ha are biased o regions wihou any arge, as are shown in Subfigure (n) and (o), will be penalized as well. In summary, boh he mean-shif algorihm and he dynamics model penalize erroneous paricle hypoheses and improve he robusness of he overall racking sysem. Figure 3 shows more racking resuls from hree differen sequences. All of hem are able o correcly mainain he ideniies of he players regardless of parial of complee occlusions.

10 116 Y. Cai, N. de Freias, and J.J. Lile (a) Frame 1 (b) Frame 30 (c) Frame 39 (d) Frame 50 (e) Frame 58 (f) Frame 30 (g) Frame 39 (h) Frame 50 (i) Frame 58 (j) Frame 30 (k) Frame 39 (l) Frame 50 (m) Frame 58 (n) Frame 30 (o) Frame 39 (p) Frame 50 (q) Frame 58 (r) Frame 30 (s) Frame 39 () Frame 50 (u) Frame 58 Fig. 2. Each row is a close-up view of he recangular region in (a). Subfigures (b-e) show he racking resuls of he sysem in [4]. Subfigures (f-i) show he racking resuls of our sysem. Subfigures (j-u) show he paricle represenaion of each arge during he racking process. Differen arges are labelled wih recangles of differen colors.

11 Robus Visual Tracking for Muliple Targes 117 (a) Frame 79 (b) Frame 83 (c) Frame 88 (d) Frame 98 (e) Frame 28 (f) Frame 34 (g) Frame 42 (h) Frame 59 (i) Frame 8 (j) Frame 12 (k) Frame 14 (l) Frame 20 Fig. 3. Each row in he figure shows he racking resuls of hree differen sequences where he op one is he same sequence as he one shown in Figure 2 7 Conclusions In his paper, we devoe our endeavors o building a racking sysem ha is able o robusly rack muliple arges and correcly mainain heir ideniies regardless of background cluer, camera moions and muual occlusion beween arges. The new paricle filer framework is more suiable for racking a variable number of arges. The recificaion echnique compensaes for he camera moion and make he moion of arges easier o predic by he second order auoregression model. The linear opimizaion algorihm achieves he global opimal soluion o correcly assign boosing deecions o he exising racks. Finally, he mean-shif embedded paricle filer is able o sabilize he rajecory of he arges and improve he dynamics model predicion. I biases paricles o new locaions wih high likelihood so ha he variance of paricle ses decreases significanly. Acknowledgemens This work has been suppored by grans from NSERC, he GEOIDE Nework of Cenres of Excellence and Honeywell Video Sysems.

12 118 Y. Cai, N. de Freias, and J.J. Lile References 1. Isard, M., Blake, A.: CONDENSATION-Condiional Densiy Propagaion for Visual Tracking. Inernainal Journal on Compuer Vision 29(1) (1998) Hue, C., Le Cadre, J., Pèrez, P.: Tracking Muliple Objecs wih Paricle Filering. In: IEEE Transacions on Aerospace and Elecronic Sysems. Volume 38. (2003) Vermaak, J., Douce, A., Pèrez, P.: Mainaining Muli-modaliy hrough Mixure Tracking. In: Inernainal Conference on Compuer Vision. Volume II. (2003) Okuma, K., Taleghani, A., de Freias, J., Lile, J., Lowe, D.: A Boosed Paricle Filer: Muliarge Deecion and Tracking. In: European Conference on Compuer Vision. Volume I. (2004) Viola, P., Jones, M.: Robus Real-Time Face Deecion. Inernainal Journal on Compuer Vision 57(2) (2004) Kang, J., Cohen, I., Medioni, G.: Soccer Player Tracking across Uncalibraed Camera Sreams. In: Join IEEE Inernaional Workshop on Visual Surveillance and Performance Evaluaion of Tracking and Surveillance (VS-PETS) In Conjuncion wih ICCV. (2003) Zhao, T., Nevaia, R.: Tracking Muliple Humans in Complex Siuaions. IEEE Transacions on Paern Analysis and Machine Inelligence 26(9) (2004) MacCormick, J., Blake, A.: A Probabilisic Exclusion Principle for Tracking Muliple Objecs. Inernainal Journal on Compuer Vision 39(1) (2000) Isard, M., MacCormick, J.: BraMBLe: A Bayesian Muliple-Blob Tracker. In: Inernainal Conference on Compuer Vision. Volume II. (2001) Rischer, J., Tu, P., Krahnsoever, N.: Simulaneous esimaion of segmenaion and shape. In: CVPR05. Volume II. (2005) Shan, C., Wei, Y., Tan, T., Ojardias, F.: Real Time Hand Tracking by Combining Paricle Filering and Mean Shif. In: Inernaional Conference on Auomaic Face and Gesure Recogniion. (2004) Comaniciu, D., Ramesh, V., Meer, P.: Kernel-based Objec Tracking. IEEE Transacions on Paern Analysis and Machine Inelligence 25(5) (2003) Pèrez, P., Hue, C., Vermaak, J., Gangne, M.: Color-Based Probabilisic Tracking. In: European Conference on Compuer Vision. Volume I. (2002) Okuma, K., Lile, J., Lowe, D.: Auomaic Recificaion of Long Image Sequences. In: Asian Conference on Compuer Vision. (2004) 15. Harley, R., Zisserman, A.: Muliple View Geomery in Compuer Vision. Cambridge Universiy Press (2000) 16. Blackman, S., Popoli, R.: Design and Analysis of Modern Tracking Sysems. Arech House, Norwood (1999) 17. Bersekas, D.: Linear Nework Opimizaion: Algorihms and Codes. The MIT Press, Cambridge (1991) 18. Comaniciu, D., Ramesh, V., Meer, P.: Real-ime racking of non-rigid objecs using mean shif. In: Inernaional Conference on Compuer Vision and Paern Recogniion. (2000) Cheng, Y.: Mean Shif, Mode Seeking, and Clusering. IEEE Transacions on Paern Analysis and Machine Inelligence 17(8) (1995) Comaniciu, D., Meer, P.: Mean Shif: A Robus Approach Toward Feaure Space Analysis. IEEE Transacions on Paern Analysis and Machine Inelligence 24(5) (2002)

Robust Visual Tracking for Multiple Targets

Robust Visual Tracking for Multiple Targets Robust Visual Tracking for Multiple Targets Yizheng Cai, Nando de Freitas, and James J. Little University of British Columbia, Vancouver, B.C., Canada, V6T 1Z4, {yizhengc, nando, little}@cs.ubc.ca Abstract.

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

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

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

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

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

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

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

Multiple View Discriminative Appearance Modeling with IMCMC for Distributed Tracking

Multiple View Discriminative Appearance Modeling with IMCMC for Distributed Tracking Muliple View Discriminaive ing wih IMCMC for Disribued Tracking Sanhoshkumar Sunderrajan, B.S. Manjunah Deparmen of Elecrical and Compuer Engineering Universiy of California, Sana Barbara {sanhosh,manj}@ece.ucsb.edu

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

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

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

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

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

Visual Indoor Localization with a Floor-Plan Map

Visual Indoor Localization with a Floor-Plan Map Visual Indoor Localizaion wih a Floor-Plan Map Hang Chu Dep. of ECE Cornell Universiy Ihaca, NY 14850 hc772@cornell.edu Absrac In his repor, a indoor localizaion mehod is presened. The mehod akes firsperson

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

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

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

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

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

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

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

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

Tracking a Large Number of Objects from Multiple Views

Tracking a Large Number of Objects from Multiple Views Tracking a Large Number of Objecs from Muliple Views Zheng Wu 1, Nickolay I. Hrisov 2, Tyson L. Hedrick 3, Thomas H. Kun 2, Margri Beke 1 1 Deparmen of Compuer Science, Boson Universiy 2 Deparmen of Biology,

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

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

Tracking a Large Number of Objects from Multiple Views

Tracking a Large Number of Objects from Multiple Views Boson Universiy Compuer Science Deparmen Technical Repor BUCS-TR 2009-005 Tracking a Large Number of Objecs from Muliple Views Zheng Wu 1, Nickolay I. Hrisov 2, Tyson L. Hedrick 3, Thomas H. Kun 2, Margri

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

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

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

Multi-camera multi-object voxel-based Monte Carlo 3D tracking strategies

Multi-camera multi-object voxel-based Monte Carlo 3D tracking strategies RESEARCH Open Access Muli-camera muli-objec voxel-based Mone Carlo 3D racking sraegies Crisian Canon-Ferrer *, Josep R Casas, Monse Pardàs and Enric Mone Absrac This aricle presens a new approach o he

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

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

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

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

Multi-Target Detection and Tracking from a Single Camera in Unmanned Aerial Vehicles (UAVs)

Multi-Target Detection and Tracking from a Single Camera in Unmanned Aerial Vehicles (UAVs) 2016 IEEE/RSJ Inernaional Conference on Inelligen Robos and Sysems (IROS) Daejeon Convenion Cener Ocober 9-14, 2016, Daejeon, Korea Muli-Targe Deecion and Tracking from a Single Camera in Unmanned Aerial

More information

An Iterative Scheme for Motion-Based Scene Segmentation

An Iterative Scheme for Motion-Based Scene Segmentation An Ieraive Scheme for Moion-Based Scene Segmenaion Alexander Bachmann and Hildegard Kuehne Deparmen for Measuremen and Conrol Insiue for Anhropomaics Universiy of Karlsruhe (H), 76 131 Karlsruhe, Germany

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

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

Real time 3D face and facial feature tracking

Real time 3D face and facial feature tracking J Real-Time Image Proc (2007) 2:35 44 DOI 10.1007/s11554-007-0032-2 ORIGINAL RESEARCH PAPER Real ime 3D face and facial feaure racking Fadi Dornaika Æ Javier Orozco Received: 23 November 2006 / Acceped:

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

Track-based and object-based occlusion for people tracking refinement in indoor surveillance

Track-based and object-based occlusion for people tracking refinement in indoor surveillance Trac-based and objec-based occlusion for people racing refinemen in indoor surveillance R. Cucchiara, C. Grana, G. Tardini Diparimeno di Ingegneria Informaica - Universiy of Modena and Reggio Emilia Via

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

Simultaneous Localization and Mapping with Stereo Vision

Simultaneous Localization and Mapping with Stereo Vision Simulaneous Localizaion and Mapping wih Sereo Vision Mahew N. Dailey Compuer Science and Informaion Managemen Asian Insiue of Technology Pahumhani, Thailand Email: mdailey@ai.ac.h Manukid Parnichkun Mecharonics

More information

IROS 2015 Workshop on On-line decision-making in multi-robot coordination (DEMUR 15)

IROS 2015 Workshop on On-line decision-making in multi-robot coordination (DEMUR 15) IROS 2015 Workshop on On-line decision-making in muli-robo coordinaion () OPTIMIZATION-BASED COOPERATIVE MULTI-ROBOT TARGET TRACKING WITH REASONING ABOUT OCCLUSIONS KAROL HAUSMAN a,, GREGORY KAHN b, SACHIN

More information

MOTION TRACKING is a fundamental capability that

MOTION TRACKING is a fundamental capability that TECHNICAL REPORT CRES-05-008, CENTER FOR ROBOTICS AND EMBEDDED SYSTEMS, UNIVERSITY OF SOUTHERN CALIFORNIA 1 Real-ime Moion Tracking from a Mobile Robo Boyoon Jung, Suden Member, IEEE, Gaurav S. Sukhame,

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 LSTM-Autoencoders for Face De-Occlusion in the Wild

Robust LSTM-Autoencoders for Face De-Occlusion in the Wild IEEE TRANSACTIONS ON IMAGE PROCESSING, DRAFT 1 Robus LSTM-Auoencoders for Face De-Occlusion in he Wild Fang Zhao, Jiashi Feng, Jian Zhao, Wenhan Yang, Shuicheng Yan arxiv:1612.08534v1 [cs.cv] 27 Dec 2016

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

Robot localization under perceptual aliasing conditions based on laser reflectivity using particle filter

Robot localization under perceptual aliasing conditions based on laser reflectivity using particle filter Robo localizaion under percepual aliasing condiions based on laser refleciviy using paricle filer DongXiang Zhang, Ryo Kurazume, Yumi Iwashia, Tsuomu Hasegawa Absrac Global localizaion, which deermines

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

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

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

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

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

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

Learning nonlinear appearance manifolds for robot localization

Learning nonlinear appearance manifolds for robot localization Learning nonlinear appearance manifolds for robo localizaion Jihun Hamm, Yuanqing Lin, and Daniel. D. Lee GRAS Lab, Deparmen of Elecrical and Sysems Engineering Universiy of ennsylvania, hiladelphia, A

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

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

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

A Novel Approach for Monocular 3D Object Tracking in Cluttered Environment

A Novel Approach for Monocular 3D Object Tracking in Cluttered Environment Inernaional Journal of Compuaional Inelligence Research ISSN 0973-1873 Volume 13, Number 5 (2017), pp. 851-864 Research India Publicaions hp://www.ripublicaion.com A Novel Approach for Monocular 3D Objec

More information

Improving Occupancy Grid FastSLAM by Integrating Navigation Sensors

Improving Occupancy Grid FastSLAM by Integrating Navigation Sensors Improving Occupancy Grid FasSLAM by Inegraing Navigaion Sensors Chrisopher Weyers Sensors Direcorae Air Force Research Laboraory Wrigh-Paerson AFB, OH 45433 Gilber Peerson Deparmen of Elecrical and Compuer

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

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

ACQUIRING high-quality and well-defined depth data. Online Temporally Consistent Indoor Depth Video Enhancement via Static Structure

ACQUIRING high-quality and well-defined depth data. Online Temporally Consistent Indoor Depth Video Enhancement via Static Structure SUBMITTED TO TRANSACTION ON IMAGE PROCESSING 1 Online Temporally Consisen Indoor Deph Video Enhancemen via Saic Srucure Lu Sheng, Suden Member, IEEE, King Ngi Ngan, Fellow, IEEE, Chern-Loon Lim and Songnan

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

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

Dynamic Depth Recovery from Multiple Synchronized Video Streams 1

Dynamic Depth Recovery from Multiple Synchronized Video Streams 1 Dynamic Deph Recoery from Muliple ynchronized Video reams Hai ao, Harpree. awhney, and Rakesh Kumar Deparmen of Compuer Engineering arnoff Corporaion Uniersiy of California a ana Cruz Washingon Road ana

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

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

Quantitative macro models feature an infinite number of periods A more realistic (?) view of time

Quantitative macro models feature an infinite number of periods A more realistic (?) view of time INFINIE-HORIZON CONSUMPION-SAVINGS MODEL SEPEMBER, Inroducion BASICS Quaniaive macro models feaure an infinie number of periods A more realisic (?) view of ime Infinie number of periods A meaphor for many

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

Open Access Research on an Improved Medical Image Enhancement Algorithm Based on P-M Model. Luo Aijing 1 and Yin Jin 2,* u = div( c u ) u

Open Access Research on an Improved Medical Image Enhancement Algorithm Based on P-M Model. Luo Aijing 1 and Yin Jin 2,* u = div( c u ) u Send Orders for Reprins o reprins@benhamscience.ae The Open Biomedical Engineering Journal, 5, 9, 9-3 9 Open Access Research on an Improved Medical Image Enhancemen Algorihm Based on P-M Model Luo Aijing

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

Dynamic Route Planning and Obstacle Avoidance Model for Unmanned Aerial Vehicles

Dynamic Route Planning and Obstacle Avoidance Model for Unmanned Aerial Vehicles Volume 116 No. 24 2017, 315-329 ISSN: 1311-8080 (prined version); ISSN: 1314-3395 (on-line version) url: hp://www.ijpam.eu ijpam.eu Dynamic Roue Planning and Obsacle Avoidance Model for Unmanned Aerial

More information

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER ABSTRACT Modern graphics cards for compuers, and especially heir graphics processing unis (GPUs), are designed for fas rendering of graphics.

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

Motor Control. 5. Control. Motor Control. Motor Control

Motor Control. 5. Control. Motor Control. Motor Control 5. Conrol In his chaper we will do: Feedback Conrol On/Off Conroller PID Conroller Moor Conrol Why use conrol a all? Correc or wrong? Supplying a cerain volage / pulsewidh will make he moor spin a a cerain

More information

Analysis of Various Types of Bugs in the Object Oriented Java Script Language Coding

Analysis of Various Types of Bugs in the Object Oriented Java Script Language Coding Indian Journal of Science and Technology, Vol 8(21), DOI: 10.17485/ijs/2015/v8i21/69958, Sepember 2015 ISSN (Prin) : 0974-6846 ISSN (Online) : 0974-5645 Analysis of Various Types of Bugs in he Objec Oriened

More information

Reinforcement Learning by Policy Improvement. Making Use of Experiences of The Other Tasks. Hajime Kimura and Shigenobu Kobayashi

Reinforcement Learning by Policy Improvement. Making Use of Experiences of The Other Tasks. Hajime Kimura and Shigenobu Kobayashi Reinforcemen Learning by Policy Improvemen Making Use of Experiences of The Oher Tasks Hajime Kimura and Shigenobu Kobayashi Tokyo Insiue of Technology, JAPAN genfe.dis.iech.ac.jp, kobayasidis.iech.ac.jp

More information

Hierarchical Recurrent Filtering for Fully Convolutional DenseNets

Hierarchical Recurrent Filtering for Fully Convolutional DenseNets Hierarchical Recurren Filering for Fully Convoluional DenseNes Jo rg Wagner1,2, Volker Fischer1, Michael Herman1 and Sven Behnke2 1- Bosch Cener for Arificial Inelligence - 71272 Renningen - Germany 2-

More information

RGB-D Object Tracking: A Particle Filter Approach on GPU

RGB-D Object Tracking: A Particle Filter Approach on GPU RGB-D Objec Tracking: A Paricle Filer Approach on GPU Changhyun Choi and Henrik I. Chrisensen Cener for Roboics & Inelligen Machines College of Compuing Georgia Insiue of Technology Alana, GA 3332, USA

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

Coded Caching with Multiple File Requests

Coded Caching with Multiple File Requests Coded Caching wih Muliple File Requess Yi-Peng Wei Sennur Ulukus Deparmen of Elecrical and Compuer Engineering Universiy of Maryland College Park, MD 20742 ypwei@umd.edu ulukus@umd.edu Absrac We sudy a

More information

Deformable Parts Correlation Filters for Robust Visual Tracking

Deformable Parts Correlation Filters for Robust Visual Tracking PAPER UNDER REVISION Deformable Pars Correlaion Filers for Robus Visual Tracking Alan Lukežič, Luka Čehovin, Member, IEEE, and Maej Krisan, Member, IEEE ha par-based models should be considered in a layered

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

SENSING using 3D technologies, structured light cameras

SENSING using 3D technologies, structured light cameras IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 39, NO. 10, OCTOBER 2017 2045 Real-Time Enhancemen of Dynamic Deph Videos wih Non-Rigid Deformaions Kassem Al Ismaeil, Suden Member,

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

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

Detection of salient objects with focused attention based on spatial and temporal coherence

Detection of salient objects with focused attention based on spatial and temporal coherence ricle Informaion Processing Technology pril 2011 Vol.56 No.10: 1055 1062 doi: 10.1007/s11434-010-4387-1 SPECIL TOPICS: Deecion of salien objecs wih focused aenion based on spaial and emporal coherence

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

Shortest Path Algorithms. Lecture I: Shortest Path Algorithms. Example. Graphs and Matrices. Setting: Dr Kieran T. Herley.

Shortest Path Algorithms. Lecture I: Shortest Path Algorithms. Example. Graphs and Matrices. Setting: Dr Kieran T. Herley. Shores Pah Algorihms Background Seing: Lecure I: Shores Pah Algorihms Dr Kieran T. Herle Deparmen of Compuer Science Universi College Cork Ocober 201 direced graph, real edge weighs Le he lengh of a pah

More information

Joint Feature Learning With Robust Local Ternary Pattern for Face Recognition

Joint Feature Learning With Robust Local Ternary Pattern for Face Recognition Join Feaure Learning Wih Robus Local Ternary Paern for Face Recogniion Yuvaraju.M 1, Shalini.S 1 Assisan Professor, Deparmen of Elecrical and Elecronics Engineering, Anna Universiy Regional Campus, Coimbaore,

More information

DETC2004/CIE VOLUME-BASED CUT-AND-PASTE EDITING FOR EARLY DESIGN PHASES

DETC2004/CIE VOLUME-BASED CUT-AND-PASTE EDITING FOR EARLY DESIGN PHASES Proceedings of DETC 04 ASME 004 Design Engineering Technical Conferences and Compuers and Informaion in Engineering Conference Sepember 8-Ocober, 004, Sal Lake Ciy, Uah USA DETC004/CIE-57676 VOLUME-BASED

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

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

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

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

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

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