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

Size: px
Start display at page:

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

Transcription

1 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, Taipei, Taiwan, R.O.C. 2 Deparmen of Compuer Science and Informaion Engineering, Naional Taiwan Universiy, Taipei, Taiwan, R.O.C. 3 Deparmen of Compuer and Informaion Science, Naional Chiao Tung Universiy, Hsinchu, Taiwan, R.O.C. Absrac In his paper, we propose a Bayesian approach o video objec segmenaion. Our mehod consiss of wo sages. In he firs sage, we pariion he video daa ino a se of 3D waershed volumes, where each waershed volume is a series of corresponding 2D image regions. These 2D image regions are obained by applying o each image frame he marker-conrolled waershed segmenaion, where he markers are exraced by firs generaing a se of iniial markers via emporal racking and hen refining he markers wih wo shrinking schemes: he ieraive adapive erosion and he verificaion agains a pre-simplified waershed segmenaion. Nex, in he second sage, we use a Markov random field o model he spaio-emporal relaionship among he 3D waershed volumes ha are obained from he firs sage. Then, he desired video objecs can be exraced by merging waershed volumes having similar moion characerisics wihin a Bayesian framework. A major advanage of his mehod is ha i can ake ino accoun he global moion informaion conained in each waershed volume. Our experimens have shown ha he proposed mehod has poenial for exracing moving objecs from a video sequence. Index Terms Video Objec Segmenaion, Waershed Segmenaion, 3D Waershed Volume, Markov Random Field. V I. INTRODUCTION ideo objec segmenaion plays an imporan role in many advanced video applicaions (such as in MPEG-4 [2] or in virual realiy), bu sill remains a challenging research opic. A popular approach o [21] video objec segmenaion is o combine a echnique for single image segmenaion wih a emporal racking procedure. Unforunaely, single image segmenaion is iself a very difficul problem (which may no be easier han video objec segmenaion). Oher echniques in [14][17] consider video sequences o be 3D signals and exend 2D mehods o hem, alhough he ime axis does no play he same role as he spaial axis. The drawback of his echnique is ha a moving objec in one frame mus overlap wih is corresponding objec in he nex frame. If he moion disance of he objec is large, he objec may become disconneced from one frame o he nex. Mos of he unsupervised segmenaion algorihms only uilize low-level feaures such as color, exure, moion, frame difference and hisogram [10][21]. However, wihou high-order informaion, semanic video objec exracion is hard o achieve. Therefore, many researches have allowed a cerain degree of human ineracion. For example, he mehods inroduced in [3][5] require some human ineracion for he iniial segmenaion of he firs image in he video. In fac, almos all he auomaic algorihms developed for exracing video objecs have some limiaions. For example, he auomaic mehod proposed in [21] can only exrac homogeneous regions, insead of complee objecs. Realizing ha here exiss no generic auomaic algorihm applicable o all kinds of video sequences, we focus on he problem of exracing video objecs having similar moion characerisic. The mehod proposed in his paper consiss of wo sages: (1) generaion of 3D waershed volumes, and (2) Bayesian merging of 3D waershed volumes. Deails of he wo sages will be 1

2 described in Secion II and Secion III. Experimenal resuls will be shown in Secion IV, and he conclusion will be given in Secion V. II. GENERATION OF 3D WATERSHED VOLUMES Waershed algorihm has been become popular echnique for image segmenaion [6][17][19]. Given a video clip, {I, 0 T}, we can regard he daa as one volume image. Our mehod firs pariions he volume image ino a se of 3D waershed volumes, where each 3D waershed volume is a series of corresponding 2D image regions. Fig. 1 shows he flowchar of our mehod for generaing 3D waershed volumes. These 2D image regions are obained by applying o each image frame he marker-conrolled waershed segmenaion described in Sep 2 of Secion II-B. The procedure for generaing 3D waershed volumes can be divided ino wo phases: iniial segmenaion and emporal racking. Deails of hese wo phases are described below. A. Iniial Segmenaion In he iniial phase, he firs frame of he video clip, I 0, is pariioned ino a se of 2D regions by applying he waershed segmenaion algorihm o he gradien image of I 0. However, he basic waershed ransformaion ends o produce over-segmenaion due o noise or local irregulariies in he gradien image. Since overly segmened regions may no be reliable enough for he nex phase of emporal racking, we adop a pre-processing mehod called opographic simplificaion o alleviae he over-segmenaion problem. In our curren implemenaion, he opographic surface of he gradien image is simplified by removing he local minima [20]. Firs, we apply a dilaion operaion wih a srucuring elemen of 2x2 pixels, i.e., le g 1 = Grad(I 0 ) B 2 2. Nex, we apply o Grad(I 0 ) a reconsrucion by erosion [18] from g 1 +h, i.e., le g 2 = φ (rec) [g 1 +h, Grad(I 0 )]. Noice ha using a larger h can eliminae more local minima. Finally, we can obain a reasonable segmenaion of I 0 by applying he basic waershed ransformaion o he simplified gradien image, g 2. In his paper, he above procedure of opographic simplificaion followed by waershed ransformaion will be referred o as he pre-simplified waershed segmenaion, and will be applied again o each subsequen frame for he purpose of refining he exraced markers, as described in Sep 1.3. Afer pre-simplified waershed segmenaion, merging of a foreground region and a background region may occasionally occur. Tha means he volume of parameer h is oo large so ha waershed regions are over-simplified. The user can selec eiher a smaller h or apply some human inervenion suppored by our sysem. Our ool allows he user o draw differen markers on some pars of he region o indicae ha hey should no be merged. Then, he marker-conrolled waershed segmenaion will be performed so ha he merged region will be spli auomaically. The operaion is qui simple for he users, and his operaion, if needed, usually is required only for he firs frame. Fig. 2 shows an example of our user inervenion ool. The edge of he ha and he background are separaed by drawing differen markers on each of hem, as shown in Fig. 2(a). Fig. 2(b) and 2(c) show he resuls afer user inervenion. B. Temporal Tracking In he second phase, our algorihm repeas he following wo seps for each subsequen frame in he video clip: (i) marker exracion, (ii) marker-conrolled waershed segmenaion. The ask of marker exracion is o exrac reliable seed regions based on he segmened regions obained from he previous frame. Given hese reliable markers, he marker-conrolled waershed segmenaion can no only accuraely exrac he boundaries of he waershed regions, bu also can deec newly emerging regions. Sep1: Marker Exracion Marker exracion is crucial o he success of he emporal racking phase and deserves some special aenion here. Our mehod 2

3 for exracing markers consiss of he following hree sub-seps: Sep 1.1: Region label propagaion by moion-based backward projecion Firs, iniial markers are obained by using backward pixel projecion based on backward moion vecors. Tha is, for each pixel p in he curren frame, we assign o he region label of he corresponding pixel in he previous frame o i. The correspondence is deermined by using he backward moion vecor m p. Here, we choose o use backward moion o avoid generaing empy and conflicing areas in he curren frame. The dense field of backward moion vecors is esimaed by using a emplae-maching algorihm ha adops adapive windows, similar o he one used in [7]. To save he compuaion ime, we firs esimae a sparse field of moion vecors a every 4 4 pixel spacing. Then, he dense pixel-wise moion vecors are compued using bilinear inerpolaion. The approximaion error can be deal wih he following process. Sep 1.2: Removing unreliable pixels from iniial markers by ieraive adapive erosion Since moion vecors are usually no very accurae, we mus remove unreliable region assignmens due o erroneous pixel correspondences. In order o reduce he possibiliy of generaing false boundaries in he nex sub-sep, he exraced markers should be as large as possible, and compleely conained in heir rue corresponding regions - which are unforunaely unknown o he compuer. if Consider an iniial marker M i. A pixel p M i, is regarded as an unreliable pixel if i has an unreliable region propagaion, ha is, ε ( p) is greaer han k E i, where ( p) exure, including inensiy and color, beween he corresponding pixels): ε denoes he local mean of exural error cenered round pixel p (ha is, he error of 1 ε ( p) = I ( p ) I 1( p + m p ) (1) NUp p U p where U p = { p and is 8-neighbors having he same region label as p}, NU p is he number of elemens in he se Up, and Ei denoes he global mean of exural error for he whole area of marker M i : where NM i 1 E i = min max + I ( p ) I 1( p m p ), 2, 16 (2) N Mi p M i is he number of he pixels in marker Mi. The reason for consraining E i o 2 and 16 is o preven using an unreasonable large or unreasonable small hreshold. The wo number, 2 and 16, are deermined according o our experimens. In his sub-sep, we apply an ieraive adapive erosion o rim off unreliable border pixels of he iniial markers, as illusraed in Fig. 3. The adapive erosion (erode if unreliable ) is performed ieraively wih a cross-shaped srucuring elemen of 5 pixels, shown in Fig. 3(b), unil he resul becomes sable. Noice ha he adapively eroded marker shown in Fig. 3(e) is a union of he normally eroded marker (shown in Fig. 3 (d)) and he reliable pixels, coloured in whie, are conained in he border porions (shown in Fig. 3(c)). Noe ha using a lower k can eliminae more marker pixels. In he case of foreground and background objecs, which are no disincive, k should be se conservaively. We found ha k = 1.2 works well for mos MPEG-4 es sequences in hand. The resuling markers wih differen values of k using frame 116 of he foreman sequence are shown in Fig. 4. Pixels in black represen any undefined areas. Sep 1.3: Removing unreliable pixels by checking wih a pre-simplified waershed segmenaion Here, we firs generaed a reasonably fine segmenaion of he curren frame by applying he pre-simplified waershed 3

4 segmenaion described in Secion II-A, wih a small value of parameer h. For each generaed waershed region, check if i conains only one marker and he sole marker occupies more han half of he waershed region. If so, he sole major marker will be reained for driving he marker-conrolled waershed segmenaion in he nex sep. Oherwise, he marker pixel in his waershed region will be considered unreliable, and will be removed from he markers, as illusraed in Fig. 5. Fig. 6 shows he final markers obained by applying his sub-sep o he markers shown on Fig. 4. We can see ha afer his sep, small and ambiguous pieces of he marker are removed. Sep 2. Marker-conrolled waershed segmenaion Based on he reliable markers obained from he las sep, we can hen exrac more precise region boundaries by using he marker-conrolled waershed segmenaion [9][21]. One problem accompanying marker-conrolled segmenaion is ha no newly exposed regions can be exraced wihou creaing new markers. To solve his problem, we modify he marker-conrolled waershed algorihm slighly. For he flooding process of he marker-conrolled waershed algorihm used in [21], when he waer coming from wo differen basins is abou o mee, he wo basins are merged if boh have he same label or a leas one of hem is unlabeled. Our modificaion for creaing new markers is if he dynamics of an unlabeled basin larger han a cerain hreshold [11][8], he basin will be given a new label (Fig. 7). Fig. 8 shows he resul of deecing new regions using frame 26 and 27 of he coasguard sequence. The big boa is enering he image from he lef, and he background waer can be deeced as a new region. III. BAYESIAN MERGING OF WATERSHED VOLUMES Once he 3D waershed volumes are generaed, as described in Secion 2, we need o merge hem ino a se of desired video objecs. Here, we propose a Bayesian approach o merging waershed volumes having similar moion characerisics, hoping ha more global moion informaion can be uilized wihin a formal framework. Here, we use a Markov random field (MRF) o model he spaial and emporal relaionships among differen waershed volumes. A closely relaed work is he one done by Gelgon and Bouhemy, which uses region-level MRFs o rack a spaial image pariion [4]. Anoher work proposed by Paras e al. [14] labels waershed segmens by MAP. The labeling crierion is he maximizaion of he condiional a poseriori probabiliy of he labeling field given he moion hypohesis, he esimae of he label field of he previous frame, and image inensiies. However, our mehod is differen from heirs, no only in how he MRF is applied (we employ he MRF afer racking while hey do i before racking), bu also in how he class-condiional probabiliy is modeled. A. Exracion of Feaures from 3D Waershed Volume Before applying he Bayesian merging o 3D waershed volumes, he represenaive feaures for each waershed volume need o be exraced. Moion informaion is an imporan cue o produce semanic objecs. Hence, for each waershed volume v, we consruc a feaure vecor θ v based on moion informaion. We firs decompose each waershed volume v ino a se of regions { R 0 () v () v T} v b e, where denoes a region which can be obained by inersecing frame wih he waershed volume R v v, b(v) and e (v) are he indices of he beginning frame and he ending frame of he waershed volume v, respecively. Noe ha he indices of he beginning and ending frames of he waershed volumes can vary for he waershed volume v due o he appearance or disappearance of objecs in he scene. In pracical siuaions, image moion of a rigid objec can be approximaely modeled by a small number of moion parameers. If wo regions roughly correspond o he same 3D rigid objec, he moion parameer should be abou he same. From he above observaion, we compue a moion parameer vecor θ v for each region R v by applying he Leas-Median Squares (LMedS) robus esimaor [15] o he backward dense moion field obained from Sep 1.1 of Secion II-B. The moion parameers can be 4

5 esimaed by θˆ v = arg min median m ( ; p u p θ v ) θ v p Rv (3) where u(.) is a parameerized moion field, is defined as wo-norm, and m p is he moion vecor of pixel p in frame. Afer he parameers for all he regions in he waershed volume v are deermined, we can consruc a moion feaure vecor: [ b() v, b() v + 1,..., () v θv = θv θv θ e v ]. Noice ha he dimensionaliy of θv is (e(v)- b (v)+1) d, where d is he dimension of θ. In our curren implemenaion, he moion characerisics of R v are described by a consan moion field, ha is, u( p; θ v ) = θ v, where θ v = [ m x, m y ] and m x and m y are he coordinaes of he mean moion vecor. If an objec undergoes a complex moion or deformaion, a more complex moion model, such as a six-parameer affine model or eigh-parameer quadraic model, should be used o enhance discriminaive abiliy [12]. Once a complex moion model, such as a six-parameer affine model, is adoped, he equaions presened in nex secion should be modified slighly. B. The Proposed Mehod In his work, we assume ha he number of video objecs, N, o be exraced (including he background objecs) is known. Given a se of 3D waershed volumes V = {v j, j=1,,k}, where K is he number of 3D waershed volumes, a Volume Adjacency Graph (VAG) can be consruced o express he neighborhood relaionship among 3D waershed volumes. Each node in he graph corresponds o a waershed volume, and beween wo volumes exiss an arc if he volumes are spaially conneced. Nex, we define { v } { } a label field L = l lv [1.. N], v V on he VAG. Given M = θ v v V, we esimae he labeling field L by maximizing he a poseriori probabiliy (MAP). Using he Bayes rule, he a poseriori probabiliy densiy funcion can be expressed as: P ( L M ) P( M L) P( L) (4) The firs erm on he righ-hand side of (4) is he condiional probabiliy disribuion P(M L). I is modeled as a Gaussian disribuion, which implies ha each objec should have minimum moion variance. ( ) 1 e( v) P 2 v µ v V 2σ l = b ( v) ( ) M L exp θ ( l ) where l is he mean of he parameer vecors of all waershed volumes in frame whose corresponding labels are l, µ v v funcion of he size of he video objec. v 2 (5) σ l is a The second erm on he righ-hand side of (4) is he prior probabiliy disribuion P(L), which is a regularizaion erm. To ake ino accoun he degree of adjacency beween wo waershed volumes, we direcly exend a measure of adjacency degree beween wo regions proposed in [4] o ha beween wo waershed volumes: where l v j, v k l v j, vk b( v j, vk ) = l v v + g j g j, k k is he area of he common border beween vi and v j, and g j and g k are he graviy ceners of v j and v k, respecively. We model he prior as a Gibbs disribuion. Before defining a Gibbs disribuion, we need o define he cliques. Here, only wo-sie cliques are considered and sraighforwardly obained from he arc of he VAG. Le C v be he se of all binary cliques. The Gibbs disribuion is given by (6) 5

6 where Z b is a normalizing consan and U b (L), he regularizaion poenial, is defined as 1 P( L) = exp( U b ( L) ) (7) Zb U ( L) = b( v, v ) δ ( l, l ) (8) b ( v j, vk ) C v where δ(.) is a Kronecker dela funcion. The regularizaion erm ends o favor idenical labels for wo neighboring volume sies. The maximum a poseriori probabiliy (MAP) esimae of L is obained by minimizing he following energy funcion: j k 1 e( v) 2 Lˆ = arg min ( ) θ + 2 v µ lv b( v j, vk ) δ ( lv, l ) j vk (9) L v V 2σ = b( v) ( v j, vk ) C v Energy minimizaion is performed using an ICM algorihm proposed by Besag [1], someimes also called he greedy algorihm. A each ieraion, each volume sies is visied. The label of each sie is eiher changed o he label ha yields maximal decrease of he energy funcion, or lef unchanged if no energy reducion is possible. The process sops when no more changes can be made. The iniializaion of he label L is esimaed by he K-means algorihm. The iniial cluser means { µ (l ) 1 l N} for he K-means algorihm are esimaed as follows. The firs cluser mean is he mean of he oal moion parameers. Tha is, v j vk () = T 1 µ (1) ( µ 0 (1), µ 1 (1),.., µ (1)); µ l = θv (10) N v V where N is he number of he waershed volume v inersecing wih frame. The c h -cluser mean is he feaure vecor he farhes disance from he neares cluser mean θ v ha has () v θv () v () c = 1 e () () + µ arg max min c c v V v e v b v µ θ 1 1 1= b In summary, he algorihm of our mehod for merging waershed volumes ino video objecs can be described as follows: Inpu: Volume Adjacency Graph (VAG), K 1. Obain iniial cluser means for K-means algorihm using equaions (10) and (11). 2. Obain iniial label for each waershed volume by applying K-means algorihm. 3. Updae labels for all volumes by applying ICM algorihm based on equaion (9). Oupu: Labels of all waershed volumes ( c ) (11) IV. EXPERIMENTAL RESULTS In his secion, we use he foreman, and coasguard sequences, shown in Fig. 9, and Fig. 10, respecively, o demonsrae he performance of our algorihm. In our curren implemenaion, he gradien images are compued on a weighed YUV colour space, i.e. w y Y+w u U+w v V. The weighing facors, w y, w u, and w v, are se o one, wo, and wo, respecively, o sress he color componens. The experimens are run in AMD Ahlon 1.2GHz PC wih 384MB RAM. The sizes of he foreman sequence and he coasguard sequence are 352x288 and 352x240. The oal execuion ime of he foreman sequence (100 frames) is 483sec, and he coasguard sequence (50 frames) is 131sec. For he experimenal resuls presened in his paper, no user inervenion has been used. However, one can always find some video sequences ha conain complex enough scene, such ha user inervenion may become necessary. 6

7 In he foreman sequence, he human body has a moderae moion and he camera is moving as well. I can be seen from Fig. 9(b), where cross-secions of waershed volumes are shown, ha he resuls obained by marker-conrolled emporal racking look prey good. By seing N = 2 (i.e., he number of video objecs o be exraced is 2), he waershed volumes depiced in Fig. 9(b) can be correcly merged ino wo video objecs: he foreman and he background, as shown in Fig. 9(c). In his sequence, we have found ha he similariy beween he moions of he head and he shoulder could be more easily deeced when considering a longer sequence. Therefore, our mehod can obain beer segmenaion resuls han hose obained by Moschni e al. [10]. In he coasguard sequence, he horizonal camera drif is presen while wo boas are moving wih differen velociies and direcions. Noice ha he bigger boa is enering he image from he lef, and is new emerging regions can be successfully exraced, as shown in Fig. 10(b). If we se N = 4, he proposed Bayesian mehod can pariion he video clip ino four differen objecs: he bigger boa, he smaller boa, he waer and he shore, as shown in Fig. 10(c). Compared wih he resuls using he mehod proposed by Paras e al.[14], he segmened boundaries we exraced are much closer o he objecs. V. CONCLUSION In his paper, we have proposed a new mehod for video objec segmenaion. This mehod firs pariions he video daa ino a se of 3D waershed volumes, and hen exracs video objecs by merging moion-coheren waershed volumes wihin a Bayesian framework. One major conribuion of his work is ha i models he prior informaion wih a MRF over a Volume Adjacency Graph (VAG), where each node of he VAG is a 3D waershed volume, and hence, is able o ake ino accoun he global moion informaion conained in each waershed volume. This mehod is appropriae for exracing objecs having similar moion because i can merge 3D waershed volumes having similar moion wih a Bayesian framework. Anoher conribuion is ha his paper proposes an efficien way o exrac reliable markers by shrinking wih wo schemes: he ieraive adapive erosion and he verificaion agains a pre-simplified waershed segmenaion. Experimenal resuls have shown ha he proposed mehod has poenial for exracing moving objecs from a video sequence. REFERENCES [1] J. Besag, On he saisical analysis of diry picures, J. R. Sa. Soc. B, 48(3): , [2] P. Daras, I. Kompasiaris, I. Grinias, G. Akrivas, G. Tzirias, S. Kollias and M.G. Srinzis: MPEG-4 Auhoring Tool using Moving Objec Segmenaion and Tracking in Video Shos, EURASIP J. on Applied Signal Processing, vol. 2003, no. 9, pp , Augus [3] D. Gaica-Perez, G. Gu and Ming-Ting Sun, Semanic Video Objec Exracion Using Four-Band Waershed and Pariion Laice operaors, IEEE Trans Circui Sysems Video Technology, 11(5): , [4] M. Gelgon and P. Bouhemy, A Region-Level Moion-Based Graph Represenaion and Labeling for Tracking a Spaial Image Pariion, Paern Recogniion, 30(4): , [5] C. Gu and M. Lee, Semiauomaic Segmenaion and Tracking of Semanic Video Objecs, IEEE Trans. Circui Sysems Video Technology, 8(5): , [6] K. Haris, S. N. Efsraiadis, N. Maglaveras, and A. K. Kasaggelos, Hybrid Image Segmenaion Using Waersheds and Fas Region Merging, IEEE Trans. Image Processing, 7(12): , [7] T. Kanade and M. Okuomi, A Sereo Maching Algorihm wih an Adapive Window: Theory and Experimen, IEEE Trans. Paern Analysis and Machine Inell., 16(9): , [8] C. Lemarechal, and Fjorof, Commens on Geodesic Saliency of Waershed Conours and Hierarchical Segmenaion, IEEE Trans. Paern Analysis and Machine Inell., 20(7): , [9] F. Meyer and S. Beucher, Morphological Segmenaion, J. Visual Commu. Image Represenaion, 1:21-46, [10] F. Moscheni, S. Bhaacharjee, and M. Kun, Spaioemporal Segmenaion Based on Region Merging, IEEE Trans. Paern Analysis and Machine Inell., 20(8): , [11] L. Najman, and M. Schmi, Geodesic Saliencey of Waershed Conours and hierarchical Segmenaion, IEEE Trans. Paern Analysis and Machine Inell., 18(12): , [12] H. T. Nguyen, M. Worring and A. Dev, Deecion of Moving Objecs in Video Using a Robus Moion Similariy Measure, IEEE Trans. on Image Processing, 9(1): , January

8 [13] M. Pardas, P. Salembier, 3D Morphological Segmenaion and Moion Esimaion for Image Sequences, Signal Processing, 38, pp , [14] I. Paras, E. A. Hendriks and R. L. Lagendijk, Video Segmenaion by MAP Labeling of Waershed Segmens, IEEE Trans. on Paern Analysis and Machine Inell., 23(3): , March [15] P.J. Rousseeuw, Leas Median of Squares Regression, J. Amer. Sais. Assoc., 79, pp , [16] J. B.T.M. Roerdink, and A. Meijser, The Waershed Transform: Definiions, Algorihms and Parallelizaion Sraegies, Fundamena Informaicae, 41: , [17] P. Salembier and M. Pardas, Hierarchical Morphological Segmenaion for Image Sequence Coding, IEEE Trans. Image Processing, 3(5): , Sepember [18] L. Vincen, Morphological Grayscale Reconsrucion in Image Analysis: Applicaion and Efficien Algorihms, IEEE Trans. on Image Processing, 2: , [19] L. Vincen and P. Soille, Waersheds in Digial Spaces: An Efficien Algorihm Based on Immersion Simulaions. IEEE. Trans. Paern Analysis and Machine Inell., 13(6): , [20] D. Wang, A Muliscale Gradien Algorihm for Image Segmenaion Using Waersheds. Paern Recogniion, 30(12): , [21] D. Wang, Unsupervised Video Segmenaion Based on Waersheds and Temporal Tracking, IEEE Trans. Circui Sysems Video Technology, 8(5): ,

9 Iniial Segmenaion (II.A) I 0 S 0 Temporal Tracking (II.B) (II.B Sep 1.1) I -1 I Moion Esimaion Region Label Propagaion Removing Unreliable Pixels (II.B Sep 1.2) Pre-Simplified Waershed Segmenaion (II.B Sep 1.3) Removing Unreliable Pixels Marker-Conrolled Waershed Segmenaion (II.B Sep 2) Fig. 1. Follow cha of generaing 3D waershed volumes. S (c) (d) Fig. 4. Markers exraced from frame 116 of sequence foreman wih differen he value of k afer Sep 1.2 for marker exracion. (a) k = 0.8. (b) k = 1.0. (b) k = 1.2. (b) k = 1.4. Regions where markers cover less han 50% of he area Ambiguous regions (conaining more han wo markers) (a) (b) Fig. 5. An example of Sep 1.3 for marker exracion. (a) Two differen markers are overlaid by he waershed lines obained from pre-simplified waershed segmenaion. (b) The shrunk marker afer removing he doubful porions. (a) (b) (c) Fig 2. Example of user assisance (a) (b) (c) (d) (e) (f) Fig. 3. An example of Sep 1.2 for marker exracion. (a) An iniial marker wih unreliable pixels coloured in grey. (b) A cross-shaped srucuring elemen of 5 pixels. (c) Border pixels removed wih he normal erosion. (d) Inerior pixels obained wih he normal erosion. (e) The eroded marker afer he firs ieraion of adapive erosion. (f) Afer he second ieraion (sable and sopped). (a) (c) (d) Fig. 6. Markers exraced from frame 116 of sequence foreman wih differen he value of k afer Sep 1.3 for marker exracion. (a) k = 0.8. (b) k = 1.0. (b) k = 1.2. (b) k = 1.4. (b) New region (a) (b) Marker Fig. 7. A new region is labelled if he dynamics of a cachmen basin exceeds a cerain hreshold. 9

10 (a) (b) (c) (d) Fig. 8. A demonsraion of deecing new region by using dynamic hresholding. (a) frame 26. (b) frame 27. (c) segmened resul of frame 26. (d) segmened resul of frame 27. (a) (b) (c) Fig.10. Coasguard sequence: Frame 1, 10, 20, 30, 40, 50. (a) Original Images. (b) Afer emporal racking. (c) Afer Bayesian merging. (a) (b) (c) Fig. 9. Foreman sequence: Frame 1, 20, 40, 60, 80, 100. (a) Original images. (b) Afer emporal racking. (c) Afer Bayesian merging. 10

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

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

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 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

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

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

Moving Object Detection Using MRF Model and Entropy based Adaptive Thresholding

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

More information

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

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

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

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

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

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

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

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

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

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

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

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

Research Article Auto Coloring with Enhanced Character Registration

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

More information

A 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

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 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

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

STRING DESCRIPTIONS OF DATA FOR DISPLAY*

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

More information

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

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

Image Content Representation

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

More information

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

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

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

Network management and QoS provisioning - QoS in Frame Relay. . packet switching with virtual circuit service (virtual circuits are bidirectional);

Network management and QoS provisioning - QoS in Frame Relay. . packet switching with virtual circuit service (virtual circuits are bidirectional); QoS in Frame Relay Frame relay characerisics are:. packe swiching wih virual circui service (virual circuis are bidirecional);. labels are called DLCI (Daa Link Connecion Idenifier);. for connecion is

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

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

CONTEXT MODELS FOR CRF-BASED CLASSIFICATION OF MULTITEMPORAL REMOTE SENSING DATA

CONTEXT MODELS FOR CRF-BASED CLASSIFICATION OF MULTITEMPORAL REMOTE SENSING DATA ISPRS Annals of he Phoogrammery, Remoe Sensing and Spaial Informaion Sciences, Volume I-7, 2012 XXII ISPRS Congress, 25 Augus 01 Sepember 2012, Melbourne, Ausralia CONTEXT MODELS FOR CRF-BASED CLASSIFICATION

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

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

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

Image Based Computer-Aided Manufacturing Technology

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

More information

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

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

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

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

Landmarks: A New Model for Similarity-Based Pattern Querying in Time Series Databases

Landmarks: A New Model for Similarity-Based Pattern Querying in Time Series Databases Lmarks: A New Model for Similariy-Based Paern Querying in Time Series Daabases Chang-Shing Perng Haixun Wang Sylvia R. Zhang D. So Parker perng@cs.ucla.edu hxwang@cs.ucla.edu Sylvia Zhang@cle.com so@cs.ucla.edu

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

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

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

More information

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

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

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

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

More information

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

Precise Voronoi Cell Extraction of Free-form Rational Planar Closed Curves

Precise Voronoi Cell Extraction of Free-form Rational Planar Closed Curves Precise Voronoi Cell Exracion of Free-form Raional Planar Closed Curves Iddo Hanniel, Ramanahan Muhuganapahy, Gershon Elber Deparmen of Compuer Science Technion, Israel Insiue of Technology Haifa 32000,

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

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

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

Automatic detection of flooded areas on ENVISAT/ASAR images using an object-oriented classification technique and an active contour algorithm.

Automatic detection of flooded areas on ENVISAT/ASAR images using an object-oriented classification technique and an active contour algorithm. Auomaic deecion of flooded areas on ENVISAT/ASAR images using an objec-oriened classificaion echnique and an acive conour algorihm. R. Heremans 1, A. Willekens 2, D. Borghys 1, B. Verbeeck 2, J. Valckenborgh

More information

The Impact of Product Development on the Lifecycle of Defects

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

More information

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 Framework for Applying Point Clouds Grabbed by Multi-Beam LIDAR in Perceiving the Driving Environment

A Framework for Applying Point Clouds Grabbed by Multi-Beam LIDAR in Perceiving the Driving Environment Sensors 215, 15, 21931-21956; doi:1.339/s15921931 Aricle OPEN ACCESS sensors ISSN 1424-822 www.mdpi.com/journal/sensors A Framewor for Applying Poin Clouds Grabbed by Muli-Beam LIDAR in Perceiving he Driving

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

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-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

Vision-Based Traffic Measurement System

Vision-Based Traffic Measurement System *J. M. Wang, **S. L. Chang, **Y. C. Chung, and **S. W. Chen Deparmen of Informaion and Compuer Educaion *Naional Taiwan Universiy **Naional Taiwan Normal Universiy Taipei, Taiwan schen@csie.nnu.edu.w Absrac

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

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

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

Robust Visual Tracking for Multiple Targets

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

More information

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

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

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

More information

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

High Resolution Passive Facial Performance Capture

High Resolution Passive Facial Performance Capture High Resoluion Passive Facial Performance Capure Derek Bradley1 Wolfgang Heidrich1 Tiberiu Popa1,2 Alla Sheffer1 1) Universiy of Briish Columbia 2) ETH Zu rich Figure 1: High resoluion passive facial performance

More information

Virtual Recovery of Excavated Archaeological Finds

Virtual Recovery of Excavated Archaeological Finds Virual Recovery of Excavaed Archaeological Finds Jiang Yu ZHENG, Zhong Li ZHANG*, Norihiro ABE Kyushu Insiue of Technology, Iizuka, Fukuoka 820, Japan *Museum of he Terra-Coa Warrlors and Horses, Lin Tong,

More information

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

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

More information

Parallel and Distributed Systems for Constructive Neural Network Learning*

Parallel and Distributed Systems for Constructive Neural Network Learning* Parallel and Disribued Sysems for Consrucive Neural Nework Learning* J. Flecher Z. Obradovi School of Elecrical Engineering and Compuer Science Washingon Sae Universiy Pullman WA 99164-2752 Absrac A consrucive

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

A Review on Block Matching Motion Estimation and Automata Theory based Approaches for Fractal Coding

A Review on Block Matching Motion Estimation and Automata Theory based Approaches for Fractal Coding Regular Issue A Review on Block Maching Moion Esimaion and Auomaa Theory based Approaches for Fracal Coding Shailesh D Kamble 1, Nileshsingh V Thakur 2, and Preei R Bajaj 3 1 Compuer Science & Engineering,

More information

Recovering Joint and Individual Components in Facial Data

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

More information

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

Mobile Robots Mapping

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

More information

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

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

Gender Classification of Faces Using Adaboost*

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

More information

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

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

TrackNet: Simultaneous Detection and Tracking of Multiple Objects

TrackNet: Simultaneous Detection and Tracking of Multiple Objects TrackNe: Simulaneous Deecion and Tracking of Muliple Objecs Chenge Li New York Universiy cl2840@nyu.edu Gregory Dobler New York Universiy greg.dobler@nyu.edu Yilin Song New York Universiy ys1297@nyu.edu

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

4.1 3D GEOMETRIC TRANSFORMATIONS

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

More information

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

3-D Object Modeling and Recognition for Telerobotic Manipulation

3-D Object Modeling and Recognition for Telerobotic Manipulation Research Showcase @ CMU Roboics Insiue School of Compuer Science 1995 3-D Objec Modeling and Recogniion for Teleroboic Manipulaion Andrew Johnson Parick Leger Regis Hoffman Marial Heber James Osborn Follow

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

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

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

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

More information

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

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

Scale Recovery for Monocular Visual Odometry Using Depth Estimated with Deep Convolutional Neural Fields

Scale Recovery for Monocular Visual Odometry Using Depth Estimated with Deep Convolutional Neural Fields Scale Recovery for Monocular Visual Odomery Using Deph Esimaed wih Deep Convoluional Neural Fields Xiaochuan Yin, Xiangwei Wang, Xiaoguo Du, Qijun Chen Tongji Universiy yinxiaochuan@homail.com,wangxiangwei.cpp@gmail.com,

More information

Visually Summarizing the Web using Internal Images and Keyphrases

Visually Summarizing the Web using Internal Images and Keyphrases Visually Summarizing he Web using Inernal Images and Keyphrases M.V.Gedam, S. A. Taale Deparmen of compuer engineering, PUNE Universiy Vidya Praishhan s College of Engg., India Absrac Visual summarizaion

More information

GrowCut - Interactive Multi-Label N-D Image Segmentation By Cellular Automata

GrowCut - Interactive Multi-Label N-D Image Segmentation By Cellular Automata GrowCu - Ineracive Muli-Label N-D Image Segmenaion By Cellular Auomaa Vladimir Vezhneves Vadim Konouchine Graphics and Media Laboraory Faculy of Compuaional Mahemaics and Cyberneics Moscow Sae Universiy,

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

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

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

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