3D Reconstruction of a Smooth Articulated Trajectory from a Monocular Image Sequence

Size: px
Start display at page:

Download "3D Reconstruction of a Smooth Articulated Trajectory from a Monocular Image Sequence"

Transcription

1 3D Reconstuction of a Smooth Aticulated Tajectoy fom a Monocula Image Sequence Hyun Soo Pak and Yase Sheikh Canegie Mellon Univesity 5 Fobes Ave., Pittsbugh, PA, 523 {hyunsoop,yase}@cs.cmu.edu Abstact An aticulated tajectoy is defined as a tajectoy that emains at a fixed distance with espect to a paent tajectoy. In this pape, we pesent a method to econstuct an aticulated tajectoy in thee dimensions given the two dimensional pojection of the aticulated tajectoy, the 3D paent tajectoy, and the camea pose at each time instant. This is a coe challenge in econstucting the 3D motion of aticulated stuctues such as the human body because endpoints of each limb fom aticulated tajectoies. We simultaneously apply activity-independent spatial and tempoal constaints, in the fom of fixed 3D distance to the paent tajectoy and smooth 3D motion. Thee exist two solutions that satisfy each instantaneous 2D pojection and aticulation constaint (a ay intesects a sphee at up to two locations) and we show that esolving this ambiguity by enfocing smoothness is equivalent to solving a binay quadatic pogamming poblem. A geometic analysis of the econstuction of aticulated tajectoies is also pesented and a measue of the econstuctibility of an aticulated tajectoy is poposed.. Intoduction Reconstucting a moving point in thee dimensions fom a sequence of two dimensional pojections is an ill-posed poblem; any point on the line of pojection connecting the camea s optical cente and an image measuement can be a solution. Yet, humans can effotlessly peceive depth if the 2D points coespond to aticulations of a known skeleton [2]. We study the conjectue that if 3D points move smoothly with a known aticulation stuctue, then it is possible to econstuct thei 3D locations fom thei 2D pojections without any activity-specific pio. The econstuction of an aticulated tajectoy has a fundamental ambigu- Poject page: iccv2/iccv_poject_page.html ity because thee ae two intesecting points that satisfy an aticulation constaint and an image measuement at each time instant [23]: fo a 2D tajectoy of F fames, thee ae 2 F 3D tajectoies that emain at fixed distance to a paent tajectoy. The econstuction of a smooth tajectoy without spatial constaints is also known to be fundamentally ambiguous when the camea tajectoy is smooth [26, 28]. We pesent an algoithm to econstuct a smooth aticulated tajectoy in 3D by simultaneously applying aticulation and smoothness constaints. The algoithm takes as input 2D pojections of the tajectoy, its paent tajectoy in 3D, and the camea pose at each time instant. We pesent a measue of econstuctibility of an aticulated tajectoy which chaacteizes the stability of estimation unde aticulation and smoothness constaints. Each tajectoy is paameteized by coefficients of a tajectoy basis in the spheical coodinate system to enfoce smoothness and aticulation constaints. We show that if a tajectoy is embedded in the tajectoy basis and aticulation constaints ae applied, the econstuction poblem is equivalent to a binay quadatic pogam which is known to be NP-had [5]. A numbe of algoithms exist that poduce an appoximate solution [24, 25, 3] and we use a banchand-bound method to poduce an initialization. We efine the aticulated tajectoies by minimizing epojection eo. The esults ae smooth, length peseved 3D tajectoies. We have applied ou algoithm to ecusively econstuct the 3D motion of a human given the 3D motion of its oot. Two geneal appoaches have been exploed in pio liteatue to econstuct human aticulated body motion. Data-diven appoaches use epositoies of exemplas to ovecome the ambiguity [2,9,34,35,37] and physics-based appoaches use dynamical models of the human body to fit to the image steam [, 36, 38]. Unlike these appoaches, ou appoach econstucts human motion fom puely geometic constaints. Thus, the taget motion is not confined to pe- The paent tajectoy in a skeleton hieachy is the poximal tajectoy to the oot tajectoy and the child tajectoy is the distal tajectoy.

2 2 S 3 R X 2 3 R 3 R O X X 2 O X X 2 X P P 3 P 3 P 3 P P P 2 P P P 2 2 3K P 2 P P P 2 2 3K P K P P S R (a) Fixed length tajectoy (b) Relative tajectoy (c) Tajectoy on a sphee Figue. (a) An aticulated tajectoy is defined as a tajectoy X 2 which peseves distance fom its paent tajectoy X acoss all time instances. (b) The aticulated tajectoy is tansfomed to the elative tajectoy, X 2 X, by collapsing X to the oigin. (c) The aticulated tajectoy lies on a sphee of adius. Thee ae two intesecting points at each time instant between the sphee and the ay connecting the camea s optical cente and an image measuement, which allow 2 F possible 3D tajectoies. defined activities o view points. We defe a detailed discussion of elated wok to the end of the pape and pesent the geomety of the poblem and ou poposed algoithm in the sequel. 2. Geomety of an Aticulated Tajectoy A point tajectoy in 3D without any constaint can be epesented by a seies of points: X = x. x F, y. y F, z. z F, () whee (x i, y i, z i ) is the Catesian coodinate of a point at i th time instant and F is the numbe of fames. If a tajectoy is smooth, it is known that the tajectoy can be expessed by a linea combination of a compact tajectoy basis [2], i.e., X = (Θa x, Θa y, Θa z ) (2) whee Θ is a F K matix composed of a collection of linea tajectoy basis, a is the coefficients o the paametes of a tajectoy, and K is the numbe of basis. If two tajectoies, X and X 2, ae aticulated, the distance between tajectoies emains constant acoss all time instances as shown in Figue (a), i.e., x 2 i + y 2 i + z 2 i = 2, i =,, F, (3) whee X = X 2 X is the elative tajectoy. When a pespective camea captues these two tajectoies, points on the tajectoies at the time instant ae pojected onto the camea plane. The camea epesentation[ in this pape ] is a 3 4 pojection matix, P i = KR i I3 C i whee I3, K, R i, and C i ae a 3 3 identity matix, the uppe tiangula intinsic matix, the camea otation matix, and the camea s optical cente vecto at the i th time instant, espectively. If we tansfom one of the tajectoies, X, to the oigin, O, the othe tajectoy, X 2, maps to the elative tajectoy, X, and a camea, P i, maps to the elative camea pose, P i with espect to X as shown in Figue (b). The tansfomed elative tajectoy lies on a sphee with adius. Thee ae two points intesecting the sphee and the ay connecting the camea s optical cente and an image measuement at each time instant as shown in Figue (c). All intesecting points ae candidate 3D points which the elative tajectoy passes and thus, thee ae 2 F possible elative tajectoies. The epesentation of a elative tajectoy between the aticulated tajectoies fom Equation (2) (Catesian coodinate epesentation) has to meet the additional quadatic equality constaints of Equation (3). Instead of the Catesian coodinate epesentation, we intoduce the spheical coodinate epesentation fo a elative tajectoy to contol the distance between tajectoies, explicitly, i.e., X = (Θa θ, Θa φ, ), (4) whee θ is inclination fom the z axis, φ is azimuth fom the x axis in the xy plane, and is the adius. This epesentation enables us to descibe an aticulated tajectoy pecisely because it satisfies the tempoal constaint and the length constaint simultaneously egadless of paametes by setting the adius constant explicitly. It also enfoces that all imputed points between fames satisfy the aticulation constaint while the Catesian epesentation does not. Fo a topological point of view, the econstuction fom the spheical coodinate system is the mapping of P 2F S 2K R while the econstuction fom the Catesian coodinate system is the mapping of P 2F P 3K as shown in Figue (c).

3 3. Method In this section, we pesent an algoithm fo ecoveing a tajectoy which satisfies spatial and tempoal constaints using the spheical coodinate epesentation of a elative tajectoy pesented in the pevious section. 3.. Objective Function of 3D Reconstuction Fom the spheical coodinate epesentation, we econstuct smooth aticulated tajectoies which minimize the epojection eos: F,P agmin X,, X P d (x ij, ˆx ij ), (5) i,j whee X j is the j th aticulated (o elative) tajectoy paameteized by (Θa θ,j, Θa φ,j, j ), d(, ) is the L 2 distance between two aguments, P is the numbe of aticulated points, and x ij and ˆx ij ae a 2D image measuement and a epojection of the j th point tajectoy at the i th time instant, espectively. If aticulated tajectoies ae sequentially linked, the tajectoies ae X j = f(x R ; X,, X j ), (6) whee f( ) is the fowad kinematic function that takes the oot tajectoy, X R, and all paent elative tajectoies, X,, X j, and outputs the j th tajectoy, X j, in the Catesian coodinate system. The epojection, ˆx ij is ( P i ˆx ij = X j (i) P 3 X i j (i), P2 X ) i j (i) P 3 X, (7) i j (i) whee P l i is the lth ow of the camea pojection matix at the i th time instant and X j (i) is the homogeneous epesentation of the i th point in the j th tajectoy, X j (i) Initialization of Equation (5) The objective function of Equation (5) is highly nonlinea and diect optimization falls into a local minimum. Theefoe, a good initialization of tajectoy paametes is necessay. When the paent joint position and the length between tajectoies ae known, thee ae two intesecting points between a sphee whose oigin is the paent joint position, X p, and a line connecting an image measuement and camea optical cente, C, at each time instant as shown in Figue 2(a). A point lying on the line is C + sv whee s is an unknown scala and v is the diection of the pojection, i.e.,v = R T K [ x T ] T. Then, the intesecting points ae X = C + s v, 2 X = C + s 2 v, (8) whee s,2 = vt C ± (v T C) 2 v 2 ( C 2 2 ) v 2 (9) and C = C X p. Fo each time instant, we have two candidate 3D points though which the econstucted tajectoy must pass. Acoss all time instances, thee ae 2 F possible tajectoies which satisfy the image measuements. Among those tajectoies, we look fo the tajectoy best descibed by the tajectoy basis. Let χ be the elative diection vecto with espect to the paent point as shown in Figue 2(a). Fo each time instant, χ i takes eithe χ i o 2 χ i, i.e., χ i = χ i b i + 2 χ i ( b i ), = ( χ i 2 χ i )b i + 2 χ i, whee b i {, }. () Then, all possible tajectoies can be epesented as: χ χ 2 χ. =... b +. χ F χ 2 F χ F o χ = Eb + F, () whee b is a binay vaiable vecto, χ i and 2 χ i ae two elative diection vectos, and χ i = χ i 2 χ i. Finding the best tajectoy is equivalent to finding the binay vecto, b, which minimizes the following cost, b = agmin ( ΘΘ T I ) (Eb + F) 2, (2) b subject to b {, } F. Note that ΘΘ T I is the pojection opeation onto the null space of the tajectoy basis, Θ. Equation (2) is a quadatic poblem ove binay vaiables. A binay quadatic pogamming poblem is NP-had in geneal. The stuctue of ou poblem does not fall into one of the solvable cases; ou quadatic matix has positive offdiagonal elements [3], is a non-singula matix [3, 4], and cannot be epesented by a ti-/five-diagonal matix [6]. Also, the undelying gaph stuctue is not seies paallel [6]. Thus, in theoy, this is an intactable poblem. Howeve, a numbe of appoaches have been poposed to appoximate a solution of the poblem efficiently using spectal o semidefinite elaxation. A banch-and-bound outine 2 with binay elaxation is one technique fo global optimization. Since ou quadatic matix is positive definite, the objective function behaves convexly in a banched ectangle, which enables us to define a tight lowe bound of the ectangle in polynomial time. 2 hybid/tools/miqp/

4 C x 2 X v 2 χ X P (a) Intesection χ X t =2 t = α Â l ˆB B γ γ C X (b) Relative tajectoy X 2 X Figue 2. (a) Thee ae two solutions, X and 2 X which satisfy the aticulation constaint and an image measuement. (b) Aticulated tajectoy and the camea pose ae tansfomed with espect to the paent tajectoy. Once b is ecoveed, we poject χ = Eb + F onto the tajectoy basis space of the spheical coodinate system to poduce low dimensional paametes, i.e., X = (Θa θ, Θa φ, ). This yields an accuate initialization which can be efined by nonlinea optimization of Equation (5). When the elative tajectoy, X, passes a singula point in the spheical coodinate system in the pocess of pojecting χ onto the spheical coodinate system, a discontinuity of angula tajectoy occus. Fo example, when φ passes fom ɛ > to 2π ɛ, this esults in a discontinuity of the angula tajectoy because φ is defined in the inteval [, 2π). To deal with discontinuous tajectoies, we find the best angula epesentation among all spheical epesentations of χ which peseves local continuity by allowing the domains of θ and φ to be (, ). 4. Reconstuctibility We now exploe the econstuction ambiguity of an aticulated tajectoy and analyze configuations in which the econstuction is accuate. Let X be a known paent tajectoy and X 2 be an aticulated child tajectoy which ae obseved at two time instances as shown in Figue 2(b). The gound tuth elative tajectoy between X and X 2 moves fom A to B. Â and ˆB ae imposto points that satisfy the image measuements as well as the aticulation constaint. In this section, we show that the elationship between the tue tajectoy and the imposto tajectoy inheently detemines the econstuction accuacy. We define a measue of econstuctibility of an aticulated tajectoy, η a, as a citeion to chaacteize econstuction accuacy whee η a = Θ a γ C Θ a γ X, (3) γ X = Θa γx + Θ a γ X, γ C = Θa γc + Θ a γ C, and Θ is the null space of the tajectoy basis. If the econstuctibility of an aticulated tajectoy goes to infinity, thee exists a unique solution and it coesponds to the gound tuth tajectoy. This can be poven by the following. Fo A each time instant, thee ae two intesecting points and an estimation should be one of them: ˆγ = ( b)γ X + bγ C, b = o (4) whee ˆγ is an estimated angle. Fo an estimated angula tajectoy, ˆγ = (I B) γ X + Bγ C, (5) whee B is a diagonal matix whose enties takes eithe o. The best tajectoy epesented by the tajectoy basis minimizes following: agmin Θâ ˆγ 2 (6) â,b = agmin Θâ (I B) γ X Bγ C 2 (7) = â,b agmin Θâ (I B) Θa γx BΘa γc â,b (I B) Θ a γ X BΘ a γ C 2. (8) Reconstuctibility of an aticulated tajectoy goes to infinity when Θ a γ C o Θ a γ X. Fo eithe case, B has to appoach to eliminate the esidual of the null components in Equation (8), which leads to â a γx. Fom the method of Pak et al. [28], if the camea motion is slow o stationay, thee is no way to econstuct an accuate tajectoy using the tajectoy basis because it spans the camea tajectoy well. The econstuctibility of an aticulation states that if the paent tajectoy is independent of the camea tajectoy, the tajectoy econstuction is still possible because mixed motion between the camea and the paent motions induces α motion whee α is the tajectoy of viewing angles fom a camea, α, as shown in Figue 2(b). Even when camea and paent motions ae stationay, the econstuction is possible if γ X Θ because each α is a nonlinea function of γ X, i.e., α = tan (sin γ X /(l + cos γ X )) whee l is the distance between the paent tajectoy and camea tajectoy, and thus α / Θ and γ C / Θ unless l = o l = (i.e., othogaphic pojection) as shown in Figue 2(b). Figue 3(a) shows the distibution of 3D econstuction eo with espect to econstuctibility of an aticulated tajectoy, η a, fom the CMU motion captue data 3. A tajectoy initialized by binay quadatic pogamming is the best fitted tajectoy by the tajectoy basis. When η a is high ( ), 3D econstuction eo of an aticulated tajectoy is low because the gound tuth tajectoy is well descibed by the tajectoy basis and the gound tuth tajectoy and the imposto tajectoy ae well sepaable. In contast, when η a is low ( ), ou solution conveges to the imposto tajectoy because the tajectoy basis spans the imposto tajectoy bette. 3

5 ηa 2 4 (a) Reconstuctibility.2 Eo in an aticulated tajectoy Eo in the paent tajectoy Root tajectoy eo, e/ a (b) Eo in the oot tajectoy.5 Small image measuement noise Lage image measuement noise 3D econstuction eo.8 3D econstuction eo 3D econstuction eo 3D econstuction eo Initial adius scale Small image measuement noise Lage image measuement noise (c) Initialization eo of adius Missing data (%) (d) Effect of missing data Figue 3. (a) The accuacy of the econstuction is high when ηa is geate than whee the tajectoy basis spans the gound tuth tajectoy bette than the imposto tajectoy. (b) Pefomance of ou algoithm against eo in the oot tajectoy, (c) the initialization eo of the adius, (d) amount of missing data ae illustated. 5. Results To validate ou method, we tested it with the HumanEvaII dataset, synthesized tajectoies, and the CMU motion captue data quantitatively and with eal human motion examples taken by video cameas qualitatively. We use the fist K Discete Cosine Tansfom (DCT) basis4 in ode of inceasing fequency and the numbe of basis is chosen manually to span the tajectoy well. 5.. Quantitative Evaluation We compae ou method with the state-of-at human pose estimation [4, 8, 2, 29] using the HumanEva-II dataset5. Subject S2 with camea C is used to econstuct the aticulated tajectoies. Ou method esults in 28.8mm of 3D mean eo with 7.75mm standad deviation. This eo is compaable to the eo of the state-of-at pose estimation algoithms (82mm 2.4mm). It should be noted that while all methods ely on activity specific taining data to econstuct motions, ou appoach uses only activity independent geometic constaints. We geneate synthetic 2D pespective pojections fom synthetic data and the CMU motion captue data and evaluate fo thee aspects: eo in the oot tajectoy, eo in adius of an aticulated tajectoy, and missing data. Fo evaluation of eos in the oot tajectoy and adius, we set the camea stationay and vay eo of the oot tajectoy and adius eo while the oot position is moving. Fo the evaluation of missing data, we atificially emove 2D pojections andomly. We measue 3D econstuction eo of an aticulated tajectoy by vaying the atio between the aveage distance eo of the oot tajectoy, e, and the adius of the aticulated tajectoy, a, as shown in Figue 3(b). The eo in the paent tajectoy is a lowe bound on the econstuction eo 4 Hamidi and Peal [7] have shown that the DCT povides the optimal pefomance to encode the signal unde the fist ode Makov pocesses. Ahkte et al. [2] have empiically justified its optimality on motion captue data. 5 of the aticulated tajectoy. While the vaiance of the distibution fo small oot tajectoy eo (<.2) is low, i.e., the econstuction can be done eliably, the econstuction fom high oot tajectoy eo (>.3) causes high eo in the child tajectoy as well. Fo the evaluation of the eo in adius, we measue 3D econstuction eo fo eoneous adii multiplied by scale6. Figue 3(c) illustates obustness to eoneous initialization. Even though the initial scale is small (i.e., 2 ), the 3D econstuction can be done eliably because befoe solving the binay quadatic pogamming, we adjust the adius of the sphee to intesect with the line of pojection at one point at least. When the initial scale is high (> ), howeve, the econstuction becomes uneliable because the ay intesects with the sphee at all time instances and the optimization falls into a local minimum aound a mis-estimated tajectoy. We also test with the CMU motion captue data fo the evaluation of missing data caused by occlusion o measuement failue. When thee ae missing data, ou spatial and tempoal constaints enable us to impute missing points. Fo this expeiment, we atificially intoduce length eos, image measuement noise, and oot tajectoy eo while the camea is stationay. Ou algoithm poduces an aveage eo7 of 3% fo 5% missing data as shown in Figue 3(d) Qualitative Evaluation We apply ou algoithm to econstuct human body motion in 3D fom 2D pespective pojections. Reconstuction fom a stationay camea and a moving camea ae tested and the statistical anthopometic length atio of the human body is used fo the initialization of length atio with some modifications fo accuate skeleton estimation pupose. The scale of the skeleton is oughly initialized and we manually label image measuements fo aticulated points. Figue 4(a) and Figue 4(b) show the econstuction of the juggling motion and the motion in font of a webcam, 6 Initial adius scale eo means the gound tuth. = X X / X, whee X is the gound tuth tajectoy and X is the estimated tajectoy. 7 eo

6 espectively, in 3D fom a stationay video camea. We poject the 2D oot tajectoy to the unit depth plane and use it as the 3D oot tajectoy because the depth of the oot tajectoy is undedetemined fom a stationay camea. Fo both expeiments, we use the toso as the oot. Fom the oot tajectoy, all aticulated tajectoies ae econstucted ecusively. We also apply ou method to data captued fom a moving camea to ecove the playing cad motion and the yoga motion as shown in Figue 4(c) and Figue 4(d), espectively. Both camea tajectoies ae smooth and well spanned by the tajectoy basis. Fo the econstuction of the oot tajectoy, we choose a elatively igid pat of human body though a sequence and econstuct them using the stuctue fom motion algoithm. Once elative camea poses ae estimated fom the igid pat of the human body, we estimate the similaity tansfom between the elative camea poses and the oiginal camea poses estimated by 3D static stuctue. Head and toso ae used as the oot fo playing cad motion and yoga motion, espectively. 6. Related Wok Points on objects involve spatial and tempoal constaints in geneal. Begle et al. [] intoduced a spatial constaint by assuming that a shape undegoes a defomation which can be well expessed by a linea combination of a compact shape basis. Subsequent wok has discussed ways to select the basis and bilinea optimization schemes that sepaate infomation of camea motion and stuctue [9, 33, 39]. Fo multiple igid bodies, Costeia and Kanade [3] poposed a factoization method using 2D tajectoies in an image steam. Yan and Pollefeys [4] extended this noting that thee is ank loss when two igid bodies ae aticulated. In the tempoal domain, Akhte et al. [2] poposed a linea tajectoy basis using the Discete Cosine Tansfom (DCT) by assuming that an undelying 3D tajectoy moves continuously and smoothly. With a calibated camea, an algebaic solution of tajectoy econstuction in 3D fo specific algebaic cuves (lines and conics) has been poposed by Avidan and Shashua [5] and late it was genealized by Kaminski and Teiche [22]. Pak et al. [28] poposed a linea solution fo the tajectoy econstuction. Pevious methods exploit eithe spatial constaints o tempoal constaints. In ou method, we use the spatial constaint in the fom of aticulation and the tempoal constaint in the fom of smoothness. Fo human motion in paticula, human pose estimation fom a single image by applying a spatial constaint (skeletal stuctue) was poposed by Taylo [32] (paameteization of limb lengths by a scala), by Baon and Kakadiais [7] (joint motion constaint fom the anthopometic statistics), by Paameswaan and Chellappa [27] (camea pose estimation fom head oientation and igidity of toso), and by Agawal and Tiggs [] (silhouette based egession). Human motion estimation fom an image sequence of a monocula camea has been studied as an extension of human pose estimation. Two popula appoaches have been exploed: data diven appoach and physics based appoach. Data diven appoach is to lean low dimensional subspace o latent vaiables that contol undelying human skeletal motion fully using motion captue data o annotated video data. Sidenbladh et al. [8] applied a Bayesian famewok fo 3D human pose tacking using a geneative model of the human body and a pio distibution defined by a tempoal dynamics model. Howe et al. [9] showed Baysian leaning, Choo and Fleet [2] sampled high dimensional taining space fom hybid Monte Calo method, and Utasun et al. [34] used Pinciple Coodinate Analysis (PCA) fo leaning of specific motion (e.g. walking and golfing). Like Taylo s wok [32], Wei and Chai [37] intoduced a geometic solution of motion econstuction using the bone symmetic constaint fom biomechanical data. Valmade and Lucey [35] discussed the validity of Wei and Chai [37] s wok and extended thei algoithm using a stuctue fom motion scheme. Recently, physics based appoaches have eceived attention. Bubake et al. [] have shown econstuction of a bipedal locomotion fom a dynamical model and Vondak et al. [36] have applied multibody dynamics simulation to infe the most plausible human motion in 3D. Wei and Chai [38] have built an inteactive system that integates a dynamical model to captue motion fom a video. Unlike pevious methods, ou appoach elies puely on a geometic intepetation of the aticulation constaint by paameteizing a tajectoy in a way that satisfies both spatial and tempoal constaints simultaneously. 7. Discussion In this pape, we study an aticulated tajectoy which emains at a constant distance with espect to the paent tajectoy. The elative tajectoy is a tajectoy on a sphee and thee ae 2 F tajectoies that meet the spatial constaint and image measuements. Among those tajectoies, we look fo the best tajectoy spanned by the tajectoy basis and we identify that this is equivalent to solving a binay quadatic pogamming poblem. The elative tajectoy obtained by the binay quadatic pogam is paameteized by a compact tajectoy basis in the spheical coodinate system, which satisfies spatial and tempoal constaints, simultaneously. We optimize the tajectoy by minimizing epojection eo. Reconstuction of the aticulated tajectoy is fundamentally limited by the motion induced by the camea and the paent tajectoy and we popose a measue of econstuctibility of an aticulated tajectoy, which chaacteizes the econstuction accuacy. Ou esults show that we ae able to econstuct highly aticulated human motions fom a stationay camea and a moving camea.

7 Acknowledgements The authos wish to thank Tomas Simon and the anonymous eviewes fo thei invaluable comments. This wok was suppoted by NSF gant IIS Refeences [] A. Agawal and B. Tiggs. Recoveing 3D human pose fom monocula images. PAMI, [2] I. Akhte, Y. Sheikh, S. Khan, and T. Kanade. Tajectoy space: A dual epesentation fo nonigid stuctue fom motion. PAMI, 2. 2, 5, 6 [3] K. Allemand, K. Fukuda, T. M. Liebling, and E. Steine. A polynomial case of unconstained zeo-one quadatic optimization. Mathematical Pogamming, 2. 3 [4] M. Andiluka, S. Roth, and B. Schiele. Monocula 3d pose estimation and tacking by detection. In CVPR, 2. 5 [5] S. Avidan and A. Shashua. Tajectoy tiangulation: 3D econstuction of moving points fom a monocula image sequence. PAMI, 2. 6 [6] F. Baahona. A solvable case of quadatic - pogamming. Discete Applied Mathematics, [7] C. Baon and I. A. Kakadiais. Estimating anthopomety and pose fom a single image. In CVPR, 2. 6 [8] M. Begtholdt, J. Kappes, S. Schmidt, and C. Schnö. A study of pats-based object class detection using complete gaphs. IJCV, 2. 5 [9] M. Band. A diect method fo 3D factoization of nonigid motion obseved in 2D. In CVPR, [] C. Begle, A. Hetzmann, and H. Biemann. Recoveing non-igid 3D shape fom image steams. In CVPR, [] M. A. Bubake and D. J. Fleet. The kneed walke fo human pose tacking. In CVPR, 28., 6 [2] K. Choo and D. J. Fleet. People tacking using hybid monte calo filteing. In ICCV, 25., 6 [3] J. P. Costeia and T. Kanade. A multibody factoization method fom independently moving objects. IJCV, [4] J.-A. Feez, K. Fukuda, and T. M. Liebling. Solving the fixed ank convex quadatic maximization in binay vaiables by a paallel zonotope constuction algoithm. Euopean Jounal of Opeations Reseach, [5] M. R. Gaey and D. S. Johnson. Compute and Inteactability: A guide to the theoy of NP-Completeness. Feeman, 979. [6] S. Gu. Polynomial time solvable algoithms to binay quadatic pogamming poblems with q being a ti-diagonal o five-diagonal matix. In WCSP, 2. 3 [7] M. Hamidi and J. Peal. Compaison of the cosine and fouie tansfoms of makov- signal. IEEE Tans. Acoust. Speech, Signal Pocess, [8] M. J. B. Hedvig Sidenbladh and D. J. Fleet. Stochastic tacking of 3D human figues using 2D image motion. In ECCV, 2. 6 [9] N. R. Howe, M. E. Leventon, and W. T. Feeman. Bayesian econstuction of 3D human motion fom single-camea video. In NIPS, 999., 6 [2] Z. L. Husz, A. M. Wallace, and P. R. Geen. Evaluation of a hieachical patitioned paticle filte with action pimitives. In CVPR Wokshop, [2] G. Jansson, S. S. Begstom, and W. Epstein. Peceiving Events and Objects. Lawence Elbaum, 994. [22] J. Y. Kaminski and M. Teiche. A geneal famewok fo tajectoy tiangulation. Jounal of Mathematical Imaging and Vision, [23] H.-J. Lee and Z. Chen. Detemination of 3D human body postues fom a single view. Compute Vision, Gaphics, and Image Pocessing, 985. [24] P. Mez and B. Feisleben. Geedy and local seach heuistics fo unconstained binay quadatic pogamming. Jounal of Heuistics, 22. [25] C. Olsson, A. P. Eiksson, and F. Kahl. Solving lage scale binay quadatic poblems: Spectal methods vs. semidefinite pogamming. In CVPR, 27. [26] K. E. Ozden, K. Conelis, L. V. Eycken, and L. V. Gool. Reconstucting 3D tajectoies of independently moving objects using geneic constaints. CVIU, 24. [27] V. Paameswaan and R. Chellappa. View independent human body pose estimation fom a single pespective image. In CVPR, [28] H. S. Pak, T. Siatoi, I. Matthews, and Y. Sheikh. 3D econstuction of a moving point fom a seies of 2D pojections. In ECCV, 2., 4, 6 [29] P. Peusum, S. Venkatesh, and G. West. A study on smoothing fo paticle-filteed 3D human body tacking. IJCV, 2. 5 [3] J. C. Picad and P. M. Ratliff. Minimal cost cut equivalent netwoks. Management Science, [3] S. Poljak, F. Rendl, and H. Wolkowicz. A ecipe fo semidefinite elaxation fo (,)-quadatic pgamming. Jounal of Global Optimization, 995. [32] C. Taylo. Reconstuction of aticulated objects fom point coespondences in a single uncalibated image. In CVIU, 2. 6 [33] L. Toesani, A. Hetzmann, and C. Begle. Nonigid stuctue-fom-motion: Estimating shape and motion with hieachical pios. PAMI, [34] R. Utasun, D. J. Fleet, and P. Fua. Tempoal motion models fo monocula and multiview 3-D human body tacking. CVIU, 26., 6 [35] J. Valmade and S. Lucey. Deteministic 3D human pose estimation using igid stuctue. In ECCV, 2., 6 [36] M. Vondak, L. Sigal, and O. C. Jenkins. Physic simulation fo pobabilistic motion tacking. In ECCV, 28., 6 [37] X. Wei and J. Chai. Modeling 3D human poses fom uncalibated monocula images. In ICCV, 29., 6 [38] X. Wei and J. Chai. Videomocap: Modeling physically ealistic human motion fom monocula video sequences. SIG- GRAPH, 2., 6 [39] J. Xiao, J. Chai, and T. Kanade. A closed-fom solution to non-igid shape and motion ecovey. IJCV, [4] J. Yan and M. Pollefeys. A factoization-based appoach fo aticulated nonigid shape, motion and kinematic chain ecovey fom video. PAMI, 28. 6

8 (a) Juggling motion fom a stationay camea (b) Motion in font of a webcam fom a stationay camea (c) Playing motion fom a moving camea (d) Yoga motion fom a moving camea Figue 4. (a) Juggling motion, (b) motion in font of the webcam fom a stationay camea, (c) playing cad motion, and (d) yoga motion fom a moving camea. Image measuements ae supeimposed on images in the top ow and 3D econstuction of the motion coesponding to the images ae shown fom diffeent views in the second and the thid ows. The ight-most figues summaize motion by showing whole tajectoies.

Optical Flow for Large Motion Using Gradient Technique

Optical Flow for Large Motion Using Gradient Technique SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol. 3, No. 1, June 2006, 103-113 Optical Flow fo Lage Motion Using Gadient Technique Md. Moshaof Hossain Sake 1, Kamal Bechkoum 2, K.K. Islam 1 Abstact: In this

More information

Color Correction Using 3D Multiview Geometry

Color Correction Using 3D Multiview Geometry Colo Coection Using 3D Multiview Geomety Dong-Won Shin and Yo-Sung Ho Gwangju Institute of Science and Technology (GIST) 13 Cheomdan-gwagio, Buk-ku, Gwangju 500-71, Republic of Koea ABSTRACT Recently,

More information

Journal of World s Electrical Engineering and Technology J. World. Elect. Eng. Tech. 1(1): 12-16, 2012

Journal of World s Electrical Engineering and Technology J. World. Elect. Eng. Tech. 1(1): 12-16, 2012 2011, Scienceline Publication www.science-line.com Jounal of Wold s Electical Engineeing and Technology J. Wold. Elect. Eng. Tech. 1(1): 12-16, 2012 JWEET An Efficient Algoithm fo Lip Segmentation in Colo

More information

Voting-Based Grouping and Interpretation of Visual Motion

Voting-Based Grouping and Interpretation of Visual Motion Voting-Based Gouping and Intepetation of Visual Motion Micea Nicolescu Depatment of Compute Science Univesity of Nevada, Reno Reno, NV 89557 micea@cs.un.edu Géad Medioni Integated Media Systems Cente Univesity

More information

Controlled Information Maximization for SOM Knowledge Induced Learning

Controlled Information Maximization for SOM Knowledge Induced Learning 3 Int'l Conf. Atificial Intelligence ICAI'5 Contolled Infomation Maximization fo SOM Knowledge Induced Leaning Ryotao Kamimua IT Education Cente and Gaduate School of Science and Technology, Tokai Univeisity

More information

Improved Fourier-transform profilometry

Improved Fourier-transform profilometry Impoved Fouie-tansfom pofilomety Xianfu Mao, Wenjing Chen, and Xianyu Su An impoved optical geomety of the pojected-finge pofilomety technique, in which the exit pupil of the pojecting lens and the entance

More information

Segmentation of Casting Defects in X-Ray Images Based on Fractal Dimension

Segmentation of Casting Defects in X-Ray Images Based on Fractal Dimension 17th Wold Confeence on Nondestuctive Testing, 25-28 Oct 2008, Shanghai, China Segmentation of Casting Defects in X-Ray Images Based on Factal Dimension Jue WANG 1, Xiaoqin HOU 2, Yufang CAI 3 ICT Reseach

More information

Development and Analysis of a Real-Time Human Motion Tracking System

Development and Analysis of a Real-Time Human Motion Tracking System Development and Analysis of a Real-Time Human Motion Tacking System Jason P. Luck 1,2 Chistian Debunne 1 William Hoff 1 Qiang He 1 Daniel E. Small 2 1 Coloado School of Mines 2 Sandia National Labs Engineeing

More information

ISyE 4256 Industrial Robotic Applications

ISyE 4256 Industrial Robotic Applications ISyE 456 Industial Robotic Applications Quiz # Oct. 9, 998 Name This is a closed book, closed notes exam. Show wok fo poblem questions ) ( pts) Please cicle one choice fo each item. a) In an application,

More information

A Mathematical Implementation of a Global Human Walking Model with Real-Time Kinematic Personification by Boulic, Thalmann and Thalmann.

A Mathematical Implementation of a Global Human Walking Model with Real-Time Kinematic Personification by Boulic, Thalmann and Thalmann. A Mathematical Implementation of a Global Human Walking Model with Real-Time Kinematic Pesonification by Boulic, Thalmann and Thalmann. Mashall Badley National Cente fo Physical Acoustics Univesity of

More information

IP Network Design by Modified Branch Exchange Method

IP Network Design by Modified Branch Exchange Method Received: June 7, 207 98 IP Netwok Design by Modified Banch Method Kaiat Jaoenat Natchamol Sichumoenattana 2* Faculty of Engineeing at Kamphaeng Saen, Kasetsat Univesity, Thailand 2 Faculty of Management

More information

Prof. Feng Liu. Fall /17/2016

Prof. Feng Liu. Fall /17/2016 Pof. Feng Liu Fall 26 http://www.cs.pdx.edu/~fliu/couses/cs447/ /7/26 Last time Compositing NPR 3D Gaphics Toolkits Tansfomations 2 Today 3D Tansfomations The Viewing Pipeline Mid-tem: in class, Nov. 2

More information

3D Periodic Human Motion Reconstruction from 2D Motion Sequences

3D Periodic Human Motion Reconstruction from 2D Motion Sequences 3D Peiodic Human Motion Reconstuction fom D Motion Sequences Zonghua Zhang and Nikolaus F. Toje BioMotionLab, Depatment of Psychology Queen s Univesity, Canada zhang, toje@psyc.queensu.ca Abstact In this

More information

EYE DIRECTION BY STEREO IMAGE PROCESSING USING CORNEAL REFLECTION ON AN IRIS

EYE DIRECTION BY STEREO IMAGE PROCESSING USING CORNEAL REFLECTION ON AN IRIS EYE DIRECTION BY STEREO IMAGE PROCESSING USING CORNEAL REFLECTION ON AN IRIS Kumiko Tsuji Fukuoka Medical technology Teikyo Univesity 4-3-14 Shin-Katsutachi-Machi Ohmuta Fukuoka 836 Japan email: c746g@wisdomcckyushu-uacjp

More information

3D Reconstruction from 360 x 360 Mosaics 1

3D Reconstruction from 360 x 360 Mosaics 1 CENTER FOR MACHINE PERCEPTION 3D Reconstuction fom 36 x 36 Mosaics CZECH TECHNICAL UNIVERSITY {bakstein, pajdla}@cmp.felk.cvut.cz REPRINT Hynek Bakstein and Tomáš Pajdla, 3D Reconstuction fom 36 x 36 Mosaics,

More information

Assessment of Track Sequence Optimization based on Recorded Field Operations

Assessment of Track Sequence Optimization based on Recorded Field Operations Assessment of Tack Sequence Optimization based on Recoded Field Opeations Matin A. F. Jensen 1,2,*, Claus G. Søensen 1, Dionysis Bochtis 1 1 Aahus Univesity, Faculty of Science and Technology, Depatment

More information

A modal estimation based multitype sensor placement method

A modal estimation based multitype sensor placement method A modal estimation based multitype senso placement method *Xue-Yang Pei 1), Ting-Hua Yi 2) and Hong-Nan Li 3) 1),)2),3) School of Civil Engineeing, Dalian Univesity of Technology, Dalian 116023, China;

More information

Fifth Wheel Modelling and Testing

Fifth Wheel Modelling and Testing Fifth heel Modelling and Testing en Masoy Mechanical Engineeing Depatment Floida Atlantic Univesity Boca aton, FL 4 Lois Malaptias IFMA Institut Fancais De Mechanique Advancee ampus De lemont Feand Les

More information

Augmented Reality. Integrating Computer Graphics with Computer Vision Mihran Tuceryan. August 16, 1998 ICPR 98 1

Augmented Reality. Integrating Computer Graphics with Computer Vision Mihran Tuceryan. August 16, 1998 ICPR 98 1 Augmented Reality Integating Compute Gaphics with Compute Vision Mihan Tuceyan August 6, 998 ICPR 98 Definition XCombines eal and vitual wolds and objects XIt is inteactive and eal-time XThe inteaction

More information

Layered Animation using Displacement Maps

Layered Animation using Displacement Maps Layeed Animation using Displacement Maps Raymond Smith, Wei Sun, Adian Hilton and John Illingwoth Cente fo Vision, Speech and Signal Pocessing Univesity of Suey, Guildfod GU25XH, UK a.hilton@suey.ac.uk

More information

Illumination methods for optical wear detection

Illumination methods for optical wear detection Illumination methods fo optical wea detection 1 J. Zhang, 2 P.P.L.Regtien 1 VIMEC Applied Vision Technology, Coy 43, 5653 LC Eindhoven, The Nethelands Email: jianbo.zhang@gmail.com 2 Faculty Electical

More information

= dv 3V (r + a 1) 3 r 3 f(r) = 1. = ( (r + r 2

= dv 3V (r + a 1) 3 r 3 f(r) = 1. = ( (r + r 2 Random Waypoint Model in n-dimensional Space Esa Hyytiä and Joma Vitamo Netwoking Laboatoy, Helsinki Univesity of Technology, Finland Abstact The andom waypoint model (RWP) is one of the most widely used

More information

ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM

ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM ADDING REALISM TO SOURCE CHARACTERIZATION USING A GENETIC ALGORITHM Luna M. Rodiguez*, Sue Ellen Haupt, and Geoge S. Young Depatment of Meteoology and Applied Reseach Laboatoy The Pennsylvania State Univesity,

More information

This document contains the draft version of the following paper:

This document contains the draft version of the following paper: This document contains the daft vesion of the following pape: R. Sinha, S.K. Gupta, C.J. Paedis, P.K. Khosla. Extacting aticulation models fom CAD models of pats with cuved sufaces. ASME Jounal of Mechanical

More information

On Error Estimation in Runge-Kutta Methods

On Error Estimation in Runge-Kutta Methods Leonado Jounal of Sciences ISSN 1583-0233 Issue 18, Januay-June 2011 p. 1-10 On Eo Estimation in Runge-Kutta Methods Ochoche ABRAHAM 1,*, Gbolahan BOLARIN 2 1 Depatment of Infomation Technology, 2 Depatment

More information

2. PROPELLER GEOMETRY

2. PROPELLER GEOMETRY a) Fames of Refeence 2. PROPELLER GEOMETRY 10 th Intenational Towing Tank Committee (ITTC) initiated the pepaation of a dictionay and nomenclatue of ship hydodynamic tems and this wok was completed in

More information

RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES

RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES RANDOM IRREGULAR BLOCK-HIERARCHICAL NETWORKS: ALGORITHMS FOR COMPUTATION OF MAIN PROPERTIES Svetlana Avetisyan Mikayel Samvelyan* Matun Kaapetyan Yeevan State Univesity Abstact In this pape, the class

More information

Massachusetts Institute of Technology Department of Mechanical Engineering

Massachusetts Institute of Technology Department of Mechanical Engineering cm cm Poblem Massachusetts Institute of echnolog Depatment of Mechanical Engineeing. Intoduction to obotics Sample Poblems and Solutions fo the Mid-em Exam Figue shows a obotic vehicle having two poweed

More information

Adaptation of Motion Capture Data of Human Arms to a Humanoid Robot Using Optimization

Adaptation of Motion Capture Data of Human Arms to a Humanoid Robot Using Optimization ICCAS25 June 2-5, KINTEX, Gyeonggi-Do, Koea Adaptation of Motion Captue Data of Human Ams to a Humanoid Robot Using Optimization ChangHwan Kim and Doik Kim Intelligent Robotics Reseach Cente, Koea Institute

More information

Also available at ISSN (printed edn.), ISSN (electronic edn.) ARS MATHEMATICA CONTEMPORANEA 3 (2010)

Also available at  ISSN (printed edn.), ISSN (electronic edn.) ARS MATHEMATICA CONTEMPORANEA 3 (2010) Also available at http://amc.imfm.si ISSN 1855-3966 (pinted edn.), ISSN 1855-3974 (electonic edn.) ARS MATHEMATICA CONTEMPORANEA 3 (2010) 109 120 Fulleene patches I Jack E. Gave Syacuse Univesity, Depatment

More information

(a, b) x y r. For this problem, is a point in the - coordinate plane and is a positive number.

(a, b) x y r. For this problem, is a point in the - coordinate plane and is a positive number. Illustative G-C Simila cicles Alignments to Content Standads: G-C.A. Task (a, b) x y Fo this poblem, is a point in the - coodinate plane and is a positive numbe. a. Using a tanslation and a dilation, show

More information

Positioning of a robot based on binocular vision for hand / foot fusion Long Han

Positioning of a robot based on binocular vision for hand / foot fusion Long Han 2nd Intenational Confeence on Advances in Mechanical Engineeing and Industial Infomatics (AMEII 26) Positioning of a obot based on binocula vision fo hand / foot fusion Long Han Compute Science and Technology,

More information

A Shape-preserving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonuniform Fuzzification Transform

A Shape-preserving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonuniform Fuzzification Transform A Shape-peseving Affine Takagi-Sugeno Model Based on a Piecewise Constant Nonunifom Fuzzification Tansfom Felipe Fenández, Julio Gutiéez, Juan Calos Cespo and Gacián Tiviño Dep. Tecnología Fotónica, Facultad

More information

An Unsupervised Segmentation Framework For Texture Image Queries

An Unsupervised Segmentation Framework For Texture Image Queries An Unsupevised Segmentation Famewok Fo Textue Image Queies Shu-Ching Chen Distibuted Multimedia Infomation System Laboatoy School of Compute Science Floida Intenational Univesity Miami, FL 33199, USA chens@cs.fiu.edu

More information

17/5/2009. Introduction

17/5/2009. Introduction 7/5/9 Steeo Imaging Intoduction Eample of Human Vision Peception of Depth fom Left and ight eye images Diffeence in elative position of object in left and ight eyes. Depth infomation in the views?? 7/5/9

More information

Spiral Recognition Methodology and Its Application for Recognition of Chinese Bank Checks

Spiral Recognition Methodology and Its Application for Recognition of Chinese Bank Checks Spial Recognition Methodology and Its Application fo Recognition of Chinese Bank Checks Hanshen Tang 1, Emmanuel Augustin 2, Ching Y. Suen 1, Olivie Baet 2, Mohamed Cheiet 3 1 Cente fo Patten Recognition

More information

A New and Efficient 2D Collision Detection Method Based on Contact Theory Xiaolong CHENG, Jun XIAO a, Ying WANG, Qinghai MIAO, Jian XUE

A New and Efficient 2D Collision Detection Method Based on Contact Theory Xiaolong CHENG, Jun XIAO a, Ying WANG, Qinghai MIAO, Jian XUE 5th Intenational Confeence on Advanced Mateials and Compute Science (ICAMCS 2016) A New and Efficient 2D Collision Detection Method Based on Contact Theoy Xiaolong CHENG, Jun XIAO a, Ying WANG, Qinghai

More information

Motion Estimation. Yao Wang Tandon School of Engineering, New York University

Motion Estimation. Yao Wang Tandon School of Engineering, New York University Motion Estimation Yao Wang Tandon School of Engineeing, New Yok Univesity Outline 3D motion model 2-D motion model 2-D motion vs. optical flow Optical flow equation and ambiguity in motion estimation Geneal

More information

Transmission Lines Modeling Based on Vector Fitting Algorithm and RLC Active/Passive Filter Design

Transmission Lines Modeling Based on Vector Fitting Algorithm and RLC Active/Passive Filter Design Tansmission Lines Modeling Based on Vecto Fitting Algoithm and RLC Active/Passive Filte Design Ahmed Qasim Tuki a,*, Nashien Fazilah Mailah b, Mohammad Lutfi Othman c, Ahmad H. Saby d Cente fo Advanced

More information

Introduction to Medical Imaging. Cone-Beam CT. Introduction. Available cone-beam reconstruction methods: Our discussion:

Introduction to Medical Imaging. Cone-Beam CT. Introduction. Available cone-beam reconstruction methods: Our discussion: Intoduction Intoduction to Medical Imaging Cone-Beam CT Klaus Muelle Available cone-beam econstuction methods: exact appoximate Ou discussion: exact (now) appoximate (next) The Radon tansfom and its invese

More information

3D Hand Trajectory Segmentation by Curvatures and Hand Orientation for Classification through a Probabilistic Approach

3D Hand Trajectory Segmentation by Curvatures and Hand Orientation for Classification through a Probabilistic Approach 3D Hand Tajectoy Segmentation by Cuvatues and Hand Oientation fo Classification though a Pobabilistic Appoach Diego R. Faia and Joge Dias Abstact In this wok we pesent the segmentation and classification

More information

Detection and Recognition of Alert Traffic Signs

Detection and Recognition of Alert Traffic Signs Detection and Recognition of Alet Taffic Signs Chia-Hsiung Chen, Macus Chen, and Tianshi Gao 1 Stanfod Univesity Stanfod, CA 9305 {echchen, macuscc, tianshig}@stanfod.edu Abstact Taffic signs povide dives

More information

Mono Vision Based Construction of Elevation Maps in Indoor Environments

Mono Vision Based Construction of Elevation Maps in Indoor Environments 8th WSEAS Intenational onfeence on SIGNAL PROESSING, OMPUTATIONAL GEOMETRY and ARTIFIIAL VISION (ISGAV 08) Rhodes, Geece, August 0-, 008 Mono Vision Based onstuction of Elevation Maps in Indoo Envionments

More information

Obstacle Avoidance of Autonomous Mobile Robot using Stereo Vision Sensor

Obstacle Avoidance of Autonomous Mobile Robot using Stereo Vision Sensor Obstacle Avoidance of Autonomous Mobile Robot using Steeo Vision Senso Masako Kumano Akihisa Ohya Shin ichi Yuta Intelligent Robot Laboatoy Univesity of Tsukuba, Ibaaki, 35-8573 Japan E-mail: {masako,

More information

Kalman filter correction with rational non-linear functions: Application to Visual-SLAM

Kalman filter correction with rational non-linear functions: Application to Visual-SLAM 1 Kalman filte coection with ational non-linea functions: Application to Visual-SLAM Thomas Féaud, Roland Chapuis, Romuald Aufèe and Paul Checchin Clemont Univesité, Univesité Blaise Pascal, LASMEA UMR

More information

vaiation than the fome. Howeve, these methods also beak down as shadowing becomes vey signicant. As we will see, the pesented algoithm based on the il

vaiation than the fome. Howeve, these methods also beak down as shadowing becomes vey signicant. As we will see, the pesented algoithm based on the il IEEE Conf. on Compute Vision and Patten Recognition, 1998. To appea. Illumination Cones fo Recognition Unde Vaiable Lighting: Faces Athinodoos S. Geoghiades David J. Kiegman Pete N. Belhumeu Cente fo Computational

More information

Shortest Paths for a Two-Robot Rendez-Vous

Shortest Paths for a Two-Robot Rendez-Vous Shotest Paths fo a Two-Robot Rendez-Vous Eik L Wyntes Joseph S B Mitchell y Abstact In this pape, we conside an optimal motion planning poblem fo a pai of point obots in a plana envionment with polygonal

More information

Topological Characteristic of Wireless Network

Topological Characteristic of Wireless Network Topological Chaacteistic of Wieless Netwok Its Application to Node Placement Algoithm Husnu Sane Naman 1 Outline Backgound Motivation Papes and Contibutions Fist Pape Second Pape Thid Pape Futue Woks Refeences

More information

Clustering Interval-valued Data Using an Overlapped Interval Divergence

Clustering Interval-valued Data Using an Overlapped Interval Divergence Poc. of the 8th Austalasian Data Mining Confeence (AusDM'9) Clusteing Inteval-valued Data Using an Ovelapped Inteval Divegence Yongli Ren Yu-Hsn Liu Jia Rong Robet Dew School of Infomation Engineeing,

More information

Image Enhancement in the Spatial Domain. Spatial Domain

Image Enhancement in the Spatial Domain. Spatial Domain 8-- Spatial Domain Image Enhancement in the Spatial Domain What is spatial domain The space whee all pixels fom an image In spatial domain we can epesent an image by f( whee x and y ae coodinates along

More information

Improvement of First-order Takagi-Sugeno Models Using Local Uniform B-splines 1

Improvement of First-order Takagi-Sugeno Models Using Local Uniform B-splines 1 Impovement of Fist-ode Takagi-Sugeno Models Using Local Unifom B-splines Felipe Fenández, Julio Gutiéez, Gacián Tiviño and Juan Calos Cespo Dep. Tecnología Fotónica, Facultad de Infomática Univesidad Politécnica

More information

A Novel Image-Based Rendering System With A Longitudinally Aligned Camera Array

A Novel Image-Based Rendering System With A Longitudinally Aligned Camera Array EUOGAPHICS 2 / A. de Sousa, J.C. Toes Shot Pesentations A Novel Image-Based endeing System With A Longitudinally Aligned Camea Aay Jiang Li, Kun Zhou, Yong Wang and Heung-Yeung Shum Micosoft eseach, China

More information

Monte Carlo Techniques for Rendering

Monte Carlo Techniques for Rendering Monte Calo Techniques fo Rendeing CS 517 Fall 2002 Compute Science Conell Univesity Announcements No ectue on Thusday Instead, attend Steven Gotle, Havad Upson Hall B17, 4:15-5:15 (efeshments ealie) Geomety

More information

Cellular Neural Network Based PTV

Cellular Neural Network Based PTV 3th Int Symp on Applications of Lase Techniques to Fluid Mechanics Lisbon, Potugal, 6-9 June, 006 Cellula Neual Netwok Based PT Kazuo Ohmi, Achyut Sapkota : Depatment of Infomation Systems Engineeing,

More information

Monte Carlo Simulation for the ECAT HRRT using GATE

Monte Carlo Simulation for the ECAT HRRT using GATE Monte Calo Simulation fo the ECAT HRRT using GATE F. Bataille, C. Comtat, Membe, IEEE, S. Jan, and R. Tébossen Abstact The ECAT HRRT (High Resolution Reseach Tomogaph, CPS Innovations, Knoxville, TN, U.S.A.)

More information

A VECTOR PERTURBATION APPROACH TO THE GENERALIZED AIRCRAFT SPARE PARTS GROUPING PROBLEM

A VECTOR PERTURBATION APPROACH TO THE GENERALIZED AIRCRAFT SPARE PARTS GROUPING PROBLEM Accepted fo publication Intenational Jounal of Flexible Automation and Integated Manufactuing. A VECTOR PERTURBATION APPROACH TO THE GENERALIZED AIRCRAFT SPARE PARTS GROUPING PROBLEM Nagiza F. Samatova,

More information

Structured Light Stereoscopic Imaging with Dynamic Pseudo-random Patterns

Structured Light Stereoscopic Imaging with Dynamic Pseudo-random Patterns Stuctued Light Steeoscopic Imaging with Dynamic Pseudo-andom Pattens Piee Payeu and Danick Desjadins Univesity of Ottawa, SITE, 800 King Edwad, Ottawa, ON, Canada, K1N 6N5 {ppayeu,ddesjad}@site.uottawa.ca

More information

An Assessment of the Efficiency of Close-Range Photogrammetry for Developing a Photo-Based Scanning Systeminthe Shams Tabrizi Minaret in Khoy City

An Assessment of the Efficiency of Close-Range Photogrammetry for Developing a Photo-Based Scanning Systeminthe Shams Tabrizi Minaret in Khoy City Austalian Jounal of Basic and Applied Sciences, 5(1): 80-85, 011 ISSN 1991-8178 An Assessment of the Efficiency of Close-Range Photogammety fo Developing a Photo-Based Scanning Systeminthe Shams Tabizi

More information

Coordinate Systems. Ioannis Rekleitis

Coordinate Systems. Ioannis Rekleitis Coodinate Systems Ioannis ekleitis Position epesentation Position epesentation is: P p p p x y z P CS-417 Intoduction to obotics and Intelligent Systems Oientation epesentations Descibes the otation of

More information

Extracting Articulation Models from CAD Models of Parts with Curved Surfaces

Extracting Articulation Models from CAD Models of Parts with Curved Surfaces Extacting Aticulation Models fom CAD Models of Pats with Cuved Sufaces Rajaishi Sinha 1,*, Satyanda K. Gupta 2, Chistiaan J.J. Paedis 1, Padeep K. Khosla 1 1 Institute fo Complex Engineeed Systems, Canegie

More information

Frequency Domain Approach for Face Recognition Using Optical Vanderlugt Filters

Frequency Domain Approach for Face Recognition Using Optical Vanderlugt Filters Optics and Photonics Jounal, 016, 6, 94-100 Published Online August 016 in SciRes. http://www.scip.og/jounal/opj http://dx.doi.og/10.436/opj.016.68b016 Fequency Domain Appoach fo Face Recognition Using

More information

CSE 165: 3D User Interaction

CSE 165: 3D User Interaction CSE 165: 3D Use Inteaction Lectue #6: Selection Instucto: Jugen Schulze, Ph.D. 2 Announcements Homewok Assignment #2 Due Fiday, Januay 23 d at 1:00pm 3 4 Selection and Manipulation 5 Why ae Selection and

More information

Lecture # 04. Image Enhancement in Spatial Domain

Lecture # 04. Image Enhancement in Spatial Domain Digital Image Pocessing CP-7008 Lectue # 04 Image Enhancement in Spatial Domain Fall 2011 2 domains Spatial Domain : (image plane) Techniques ae based on diect manipulation of pixels in an image Fequency

More information

A Novel Automatic White Balance Method For Digital Still Cameras

A Novel Automatic White Balance Method For Digital Still Cameras A Novel Automatic White Balance Method Fo Digital Still Cameas Ching-Chih Weng 1, Home Chen 1,2, and Chiou-Shann Fuh 3 Depatment of Electical Engineeing, 2 3 Gaduate Institute of Communication Engineeing

More information

All lengths in meters. E = = 7800 kg/m 3

All lengths in meters. E = = 7800 kg/m 3 Poblem desciption In this poblem, we apply the component mode synthesis (CMS) technique to a simple beam model. 2 0.02 0.02 All lengths in metes. E = 2.07 10 11 N/m 2 = 7800 kg/m 3 The beam is a fee-fee

More information

Topic -3 Image Enhancement

Topic -3 Image Enhancement Topic -3 Image Enhancement (Pat 1) DIP: Details Digital Image Pocessing Digital Image Chaacteistics Spatial Spectal Gay-level Histogam DFT DCT Pe-Pocessing Enhancement Restoation Point Pocessing Masking

More information

Multi-azimuth Prestack Time Migration for General Anisotropic, Weakly Heterogeneous Media - Field Data Examples

Multi-azimuth Prestack Time Migration for General Anisotropic, Weakly Heterogeneous Media - Field Data Examples Multi-azimuth Pestack Time Migation fo Geneal Anisotopic, Weakly Heteogeneous Media - Field Data Examples S. Beaumont* (EOST/PGS) & W. Söllne (PGS) SUMMARY Multi-azimuth data acquisition has shown benefits

More information

UCLA Papers. Title. Permalink. Authors. Publication Date. Localized Edge Detection in Sensor Fields. https://escholarship.org/uc/item/3fj6g58j

UCLA Papers. Title. Permalink. Authors. Publication Date. Localized Edge Detection in Sensor Fields. https://escholarship.org/uc/item/3fj6g58j UCLA Papes Title Localized Edge Detection in Senso Fields Pemalink https://escholashipog/uc/item/3fj6g58j Authos K Chintalapudi Govindan Publication Date 3-- Pee eviewed escholashipog Poweed by the Califonia

More information

Dense pointclouds from combined nadir and oblique imagery by object-based semi-global multi-image matching

Dense pointclouds from combined nadir and oblique imagery by object-based semi-global multi-image matching Dense pointclouds fom combined nadi and oblique imagey by object-based semi-global multi-image matching Y X Thomas Luhmann, Folkma Bethmann & Heidi Hastedt Jade Univesity of Applied Sciences, Oldenbug,

More information

Communication vs Distributed Computation: an alternative trade-off curve

Communication vs Distributed Computation: an alternative trade-off curve Communication vs Distibuted Computation: an altenative tade-off cuve Yahya H. Ezzeldin, Mohammed amoose, Chistina Fagouli Univesity of Califonia, Los Angeles, CA 90095, USA, Email: {yahya.ezzeldin, mkamoose,

More information

POMDP: Introduction to Partially Observable Markov Decision Processes Hossein Kamalzadeh, Michael Hahsler

POMDP: Introduction to Partially Observable Markov Decision Processes Hossein Kamalzadeh, Michael Hahsler POMDP: Intoduction to Patially Obsevable Makov Decision Pocesses Hossein Kamalzadeh, Michael Hahsle 2019-01-02 The R package pomdp povides an inteface to pomdp-solve, a solve (witten in C) fo Patially

More information

The KCLBOT: Exploiting RGB-D Sensor Inputs for Navigation Environment Building and Mobile Robot Localization

The KCLBOT: Exploiting RGB-D Sensor Inputs for Navigation Environment Building and Mobile Robot Localization The KCBOT: Exploiting GB-D Senso Inputs fo Navigation Envionment Building and Mobile obot ocalization egula Pape Evangelos Geogiou 1,*, Jian Dai 1 and Michael uck 1 1 King s College ondon *Coesponding

More information

A Two-stage and Parameter-free Binarization Method for Degraded Document Images

A Two-stage and Parameter-free Binarization Method for Degraded Document Images A Two-stage and Paamete-fee Binaization Method fo Degaded Document Images Yung-Hsiang Chiu 1, Kuo-Liang Chung 1, Yong-Huai Huang 2, Wei-Ning Yang 3, Chi-Huang Liao 4 1 Depatment of Compute Science and

More information

OPTIMUM DESIGN OF 3R ORTHOGONAL MANIPULATORS CONSIDERING ITS TOPOLOGY

OPTIMUM DESIGN OF 3R ORTHOGONAL MANIPULATORS CONSIDERING ITS TOPOLOGY Copyight by ABCM Page 38 OPTIMUM DESIGN OF 3R ORTHOGONAL MANIPULATORS CONSIDERING ITS TOPOLOGY Giovana Tindade da Silva Oliveia, gtindadeso@yahoo.com.b School of Mechanical Engineeing, Fedeal Univesity

More information

A ROI Focusing Mechanism for Digital Cameras

A ROI Focusing Mechanism for Digital Cameras A ROI Focusing Mechanism fo Digital Cameas Chu-Hui Lee, Meng-Feng Lin, Chun-Ming Huang, and Chun-Wei Hsu Abstact With the development and application of digital technologies, the digital camea is moe popula

More information

Detection and tracking of ships using a stereo vision system

Detection and tracking of ships using a stereo vision system Scientific Reseach and Essays Vol. 8(7), pp. 288-303, 18 Febuay, 2013 Available online at http://www.academicjounals.og/sre DOI: 10.5897/SRE12.318 ISSN 1992-2248 2013 Academic Jounals Full Length Reseach

More information

Class 21. N -body Techniques, Part 4

Class 21. N -body Techniques, Part 4 Class. N -body Techniques, Pat Tee Codes Efficiency can be inceased by gouping paticles togethe: Neaest paticles exet geatest foces diect summation. Distant paticles exet smallest foces teat in goups.

More information

MULTI-TEMPORAL AND MULTI-SENSOR IMAGE MATCHING BASED ON LOCAL FREQUENCY INFORMATION

MULTI-TEMPORAL AND MULTI-SENSOR IMAGE MATCHING BASED ON LOCAL FREQUENCY INFORMATION Intenational Achives of the Photogammety Remote Sensing and Spatial Infomation Sciences Volume XXXIX-B3 2012 XXII ISPRS Congess 25 August 01 Septembe 2012 Melboune Austalia MULTI-TEMPORAL AND MULTI-SENSOR

More information

Extract Object Boundaries in Noisy Images using Level Set. Final Report

Extract Object Boundaries in Noisy Images using Level Set. Final Report Extact Object Boundaies in Noisy Images using Level Set by: Quming Zhou Final Repot Submitted to Pofesso Bian Evans EE381K Multidimensional Digital Signal Pocessing May 10, 003 Abstact Finding object contous

More information

Cardiac C-Arm CT. SNR Enhancement by Combining Multiple Retrospectively Motion Corrected FDK-Like Reconstructions

Cardiac C-Arm CT. SNR Enhancement by Combining Multiple Retrospectively Motion Corrected FDK-Like Reconstructions Cadiac C-Am CT SNR Enhancement by Combining Multiple Retospectively Motion Coected FDK-Like Reconstuctions M. Pümme 1, L. Wigstöm 2,3, R. Fahig 2, G. Lauitsch 4, J. Honegge 1 1 Institute of Patten Recognition,

More information

Comparisons of Transient Analytical Methods for Determining Hydraulic Conductivity Using Disc Permeameters

Comparisons of Transient Analytical Methods for Determining Hydraulic Conductivity Using Disc Permeameters Compaisons of Tansient Analytical Methods fo Detemining Hydaulic Conductivity Using Disc Pemeametes 1,,3 Cook, F.J. 1 CSRO Land and Wate, ndoooopilly, Queensland The Univesity of Queensland, St Lucia,

More information

Scaling Location-based Services with Dynamically Composed Location Index

Scaling Location-based Services with Dynamically Composed Location Index Scaling Location-based Sevices with Dynamically Composed Location Index Bhuvan Bamba, Sangeetha Seshadi and Ling Liu Distibuted Data Intensive Systems Laboatoy (DiSL) College of Computing, Geogia Institute

More information

Generalized Grey Target Decision Method Based on Decision Makers Indifference Attribute Value Preferences

Generalized Grey Target Decision Method Based on Decision Makers Indifference Attribute Value Preferences Ameican Jounal of ata ining and Knowledge iscovey 27; 2(4): 2-8 http://www.sciencepublishinggoup.com//admkd doi:.648/.admkd.2724.2 Genealized Gey Taget ecision ethod Based on ecision akes Indiffeence Attibute

More information

Desired Attitude Angles Design Based on Optimization for Side Window Detection of Kinetic Interceptor *

Desired Attitude Angles Design Based on Optimization for Side Window Detection of Kinetic Interceptor * Poceedings of the 7 th Chinese Contol Confeence July 6-8, 008, Kunming,Yunnan, China Desied Attitude Angles Design Based on Optimization fo Side Window Detection of Kinetic Intecepto * Zhu Bo, Quan Quan,

More information

Research Article. Regularization Rotational motion image Blur Restoration

Research Article. Regularization Rotational motion image Blur Restoration Available online www.jocp.com Jounal of Chemical and Phamaceutical Reseach, 6, 8(6):47-476 Reseach Aticle ISSN : 975-7384 CODEN(USA) : JCPRC5 Regulaization Rotational motion image Blu Restoation Zhen Chen

More information

A Minutiae-based Fingerprint Matching Algorithm Using Phase Correlation

A Minutiae-based Fingerprint Matching Algorithm Using Phase Correlation A Minutiae-based Fingepint Matching Algoithm Using Phase Coelation Autho Chen, Weiping, Gao, Yongsheng Published 2007 Confeence Title Digital Image Computing: Techniques and Applications DOI https://doi.og/10.1109/dicta.2007.4426801

More information

2D Transformations. Why Transformations. Translation 4/17/2009

2D Transformations. Why Transformations. Translation 4/17/2009 4/7/9 D Tansfomations Wh Tansfomations Coodinate sstem tansfomations Placing objects in the wold Move/animate the camea fo navigation Dawing hieachical chaactes Animation Tanslation + d 5,4 + d,3 d 4,

More information

Towards Adaptive Information Merging Using Selected XML Fragments

Towards Adaptive Information Merging Using Selected XML Fragments Towads Adaptive Infomation Meging Using Selected XML Fagments Ho-Lam Lau and Wilfed Ng Depatment of Compute Science and Engineeing, The Hong Kong Univesity of Science and Technology, Hong Kong {lauhl,

More information

View Synthesis using Depth Map for 3D Video

View Synthesis using Depth Map for 3D Video View Synthesis using Depth Map fo 3D Video Cheon Lee and Yo-Sung Ho Gwangju Institute of Science and Technology (GIST) 1 Oyong-dong, Buk-gu, Gwangju, 500-712, Republic of Koea E-mail: {leecheon, hoyo}@gist.ac.k

More information

Computer Graphics and Animation 3-Viewing

Computer Graphics and Animation 3-Viewing Compute Gaphics and Animation 3-Viewing Pof. D. Chales A. Wüthich, Fakultät Medien, Medieninfomatik Bauhaus-Univesität Weima caw AT medien.uni-weima.de Ma 5 Chales A. Wüthich Viewing Hee: Viewing in 3D

More information

Adaptive Exploration of Locally Satisfiable Proposals for Design Layout

Adaptive Exploration of Locally Satisfiable Proposals for Design Layout Adaptive Exploation of Locally Satisfiable Poposals fo Design Layout Figue 1: Rule-based design layout algoithms can solve eveything fom laying out domino tiles, though aanging funitue in a oom to supeimposing

More information

Real-Time Speech-Driven Face Animation. Pengyu Hong, Zhen Wen, Tom Huang. Beckman Institute for Advanced Science and Technology

Real-Time Speech-Driven Face Animation. Pengyu Hong, Zhen Wen, Tom Huang. Beckman Institute for Advanced Science and Technology Real-Time Speech-Diven Face Animation Pengyu Hong, Zhen Wen, Tom Huang Beckman Institute fo Advanced Science and Technology Univesity of Illinois at Ubana-Champaign, Ubana, IL 61801, USA Abstact This chapte

More information

Effects of Model Complexity on Generalization Performance of Convolutional Neural Networks

Effects of Model Complexity on Generalization Performance of Convolutional Neural Networks Effects of Model Complexity on Genealization Pefomance of Convolutional Neual Netwoks Tae-Jun Kim 1, Dongsu Zhang 2, and Joon Shik Kim 3 1 Seoul National Univesity, Seoul 151-742, Koea, E-mail: tjkim@bi.snu.ac.k

More information

Goal. Rendering Complex Scenes on Mobile Terminals or on the web. Rendering on Mobile Terminals. Rendering on Mobile Terminals. Walking through images

Goal. Rendering Complex Scenes on Mobile Terminals or on the web. Rendering on Mobile Terminals. Rendering on Mobile Terminals. Walking through images Goal Walking though s -------------------------------------------- Kadi Bouatouch IRISA Univesité de Rennes I, Fance Rendeing Comple Scenes on Mobile Teminals o on the web Rendeing on Mobile Teminals Rendeing

More information

Point-Biserial Correlation Analysis of Fuzzy Attributes

Point-Biserial Correlation Analysis of Fuzzy Attributes Appl Math Inf Sci 6 No S pp 439S-444S (0 Applied Mathematics & Infomation Sciences An Intenational Jounal @ 0 NSP Natual Sciences Publishing o Point-iseial oelation Analysis of Fuzzy Attibutes Hao-En hueh

More information

Fast quality-guided flood-fill phase unwrapping algorithm for three-dimensional fringe pattern profilometry

Fast quality-guided flood-fill phase unwrapping algorithm for three-dimensional fringe pattern profilometry Univesity of Wollongong Reseach Online Faculty of Infomatics - Papes (Achive) Faculty of Engineeing and Infomation Sciences 2010 Fast quality-guided flood-fill phase unwapping algoithm fo thee-dimensional

More information

Lecture 27: Voronoi Diagrams

Lecture 27: Voronoi Diagrams We say that two points u, v Y ae in the same connected component of Y if thee is a path in R N fom u to v such that all the points along the path ae in the set Y. (Thee ae two connected components in the

More information

Tissue Classification Based on 3D Local Intensity Structures for Volume Rendering

Tissue Classification Based on 3D Local Intensity Structures for Volume Rendering 160 IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, VOL. 6, NO., APRIL-JUNE 000 Tissue Classification Based on 3D Local Intensity Stuctues fo Volume Rendeing Yoshinobu Sato, Membe, IEEE, Cal-Fedik

More information

Extended Perspective Shadow Maps (XPSM) Vladislav Gusev, ,

Extended Perspective Shadow Maps (XPSM)   Vladislav Gusev, , Extended Pespective Shadow Maps (XPSM) http://xpsm.og Vladislav Gusev,.8.27, xmvlad@gmail.com Figue : XPSM esults (~4 objects in a scene, 536x536 shadow map). Intoduction Shadows ae one of the most impotant

More information

DEADLOCK AVOIDANCE IN BATCH PROCESSES. M. Tittus K. Åkesson

DEADLOCK AVOIDANCE IN BATCH PROCESSES. M. Tittus K. Åkesson DEADLOCK AVOIDANCE IN BATCH PROCESSES M. Tittus K. Åkesson Univesity College Boås, Sweden, e-mail: Michael.Tittus@hb.se Chalmes Univesity of Technology, Gothenbug, Sweden, e-mail: ka@s2.chalmes.se Abstact:

More information