#### DMCA

## Robust visual tracking for multiple targets (2006)

### Cached

### Download Links

Venue: | IN ECCV |

Citations: | 48 - 5 self |

### Citations

8751 | Distinctive image features from scale-invariant keypoints. IJCV,
- Lowe
- 2004
(Show Context)
Citation Context ...t of local features, which are unique and distinguish the object from others. The Harris corners [HS88], the Kanade-Lucas-Tomasi (KLT) features [TK91] and the Scale Invariant Feature Transform (SIFT) =-=[Low04]-=- are three typical examples. These features can be applied to specific object recognition and tracking, image matching and rectification, 3D structure reconstruction from 2D images for scene or motion... |

4683 |
Multiple view geometry in computer vision
- Hartley, Zisserman
- 2003
(Show Context)
Citation Context ...ical law of inertia, the motions of the players in hockey games are better predicted with a constant velocity autoregressive model. 3.1 Rectification Homography is defined by Hartley and Zisserman in =-=[15]-=- as an invertible mapping h between two planes. Images recorded by cameras are 2D projections of the real world. For any plane in the world, its images from a camera, which can pan, tilt, zoom or even... |

2418 | A combined corner and edge detector
- Harris, Stephens
- 1988
(Show Context)
Citation Context ...he most important characteristics. One popular approach is to represent an object or an image 8sby a set of local features, which are unique and distinguish the object from others. The Harris corners =-=[HS88]-=-, the Kanade-Lucas-Tomasi (KLT) features [TK91] and the Scale Invariant Feature Transform (SIFT) [Low04] are three typical examples. These features can be applied to specific object recognition and tr... |

2352 | Mean shift: a robust approach toward feature space analysis
- Comaniciu, Meer
- 2002
(Show Context)
Citation Context ...arametric statistical method that seeks the mode of a density distribution in an iterative procedure. It was first generalized and analyzed by Cheng in [19] and later developed by Comaniciu et al. in =-=[20]-=-. The objective of the mean-shift algorithm is to iteratively shift the current location x to a new location x ′ according to the following relation x ′ �M i=1 = aiw(ai)k ��� ai−x � � h 2� �M i=1 w(ai... |

1959 | A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking
- Arulampalam, Maskell, et al.
(Show Context)
Citation Context ...arest neighbor data association technique is used to correctly associate boosting detections with the existing tracks. Finally, the mean-shift algorithm is embedded into the particle filter framework =-=[AMGC01]-=- to stabilize the trajectories of the targets. By assembling all the contributions, the new multi-target tracking system is capable of achieving the goal stated in Section 1.2. The details of all the ... |

1825 | Robust Real-Time Face Detection,
- Viola, Jones
- 2004
(Show Context)
Citation Context ...sible that none of the observations belongs to the track. Hence, the constraints are formalized as and the solution is xi ′ j = � n i=1 xij =1,∀j � m j=1 xij ≥ 0,∀i � 1ifi ′ =argi min aij 0 otherwise =-=(5)-=- ∀j (6)s112 Y. Cai, N. de Freitas, and J.J. Little 5 Mean-Shift Embedded Particle Filter The motivation of embedding the mean-shift algorithm into the particle filter framework of our tracking system ... |

1185 | Neural network-based face detection - Rowley, Baluja, et al. - 1998 |

1037 |
Optimal filtering
- Anderson, Moore
- 1979
(Show Context)
Citation Context ...Figure 2.1: Graphical model representation of the relationship between an observed variable yt and a latent variable xt 2.2.1 Filtering In the field of object tracking, Bayesian Sequential Estimation =-=[AM79]-=-, which is also called Bayesian Filtering, is the most widely accepted framework. In this section, we present the fundamental theory of Bayesian filtering as a preparation for its sequential Monte Car... |

885 | P.: “Kernel-based object tracking
- Comaniciu, Ramesh, et al.
(Show Context)
Citation Context ...uccessful numerical approximation technique for Bayesian sequential estimation with non-linear, non-Gaussian models. In our application, the fast motion of hockey players and the color model we adopt =-=[12, 13]-=- is highly non-linear and non-Gaussian. Therefore, particle filtering is the ideal model to be the basic skeleton of our tracking system. The basic Bayesian filtering is a recursive process in which e... |

826 | Detecting Faces in Images: A Survey,
- Yang, Kriegman, et al.
- 2002
(Show Context)
Citation Context ... first class, the representation should capture the common features among all the objects in the same group. As it is beyond the scope of this thesis, we suggest readers refer to other related works. =-=[YKA02]-=- is one such reference that summarizes the approaches in face detection. However, some approaches also apply to the representation of other objects. In our system, hockey players are detected by a boo... |

805 | Real-time tracking of non-rigid objects using mean shift
- Comaniciu, Ramesh, et al.
- 2000
(Show Context)
Citation Context ...or Model We adopted the color model in [13, 4] in our application because it is successful in tracking non-rigid objects with partial occlusion. The model is originally introduced by Comaniciu et al. =-=[18]-=- for the mean-shift based object tracking. The observation of the target is represented by an N-bin color histogram extracted from the region R(xt) centered at the location xt. Itisdenotedas Q(xt)={q(... |

637 |
Design and Analysis of Modern Tracking Systems.
- Blackman, Popoli
- 1999
(Show Context)
Citation Context ...mizes the objective function C = �n �m i=1 j=1 aijxij subject to some linear constrains. Linear programming was initially used to solve this problem. Later on, it was found that the auction algorithm =-=[16]-=- is the most efficient method so far to reach the optimal solution or sub-optimal one without any practical difference. The extended auction algorithm [17] is able to solve the rectangular matrix prob... |

617 | Mean shift, mode seeking, and clustering
- Cheng
- 1995
(Show Context)
Citation Context ...xt,x0) . (9) 5.2 Mean-Shift Mean-shift is a nonparametric statistical method that seeks the mode of a density distribution in an iterative procedure. It was first generalized and analyzed by Cheng in =-=[19]-=- and later developed by Comaniciu et al. in [20]. The objective of the mean-shift algorithm is to iteratively shift the current location x to a new location x ′ according to the following relation x ′... |

583 | An algorithm for tracking multiple targets
- Reid
- 1979
(Show Context)
Citation Context ...rvation-to-track assignment and Ti are tracks that belong to hypothesis Hj. It can be evaluated using the likelihood p(ym|xn), or any other variations. 16 LTisCox’s implementation of Reid’s algorithm =-=[Rei79]-=- is a track oriented solution to MHT. Instead of maintaining a set of hypotheses, it creates a set of tracks. Hypotheses are generated at each step after the tracks are created. Pruning is performed b... |

566 | M.J.: Detecting pedestrians using patterns of motion and appearance. In: - Viola, Jones - 2003 |

514 |
Estimating uncertain spatial relationships in robotics
- Smith, Self, et al.
- 1990
(Show Context)
Citation Context ...e nonlinear functions in Equation 2.1 and Equation 2.2 for the approximation. The EKF has been widely used in the field of robotics in the last decade [DNDW + 00], especially in the sub-field of SLAM =-=[SSC90]-=-. The Unscented Kalman Filter (UKF) [vdMAdFW00] has even higher accuracy. It approximates the posterior p(xt|y1:t−1) directly instead of approximating the nonlinear functions f(·) and g(·). In particu... |

497 | Recognizing Action at a Distance
- Efros, Berg, et al.
- 2003
(Show Context)
Citation Context ...ions have already been investigated on the soccer games [VDP03, KCM03] and hockey games [OTdF + 04]. Recently, research on more higher level of motion analysis has also been carried out. Efros et al. =-=[EBMM03]-=- focus on the character of the motion of soccer players (i.e., walking, running, standing) rather than their translational motion and assume that the targets are stably tracked so that the targets are... |

484 |
Condensation—conditional density propagation for visual tracking
- Isard, Blake
- 1998
(Show Context)
Citation Context ...In this work, we address the problem of robust multi-target tracking within the application of hockey player tracking. Particle filtering was first introduced to visual tracking by Isard and Blake in =-=[1]-=-. Pérez et al. [2, 3] extended the particle filter framework to track multiple targets. Okuma et al. [4] further extended it [3] by incorporating a boosting detector [5] into the particle filter for a... |

391 | A general framework for object detection - Papageorgiou, Oren, et al. - 1998 |

352 | Color-based probabilistic tracking
- Pérez, Hue, et al.
(Show Context)
Citation Context ...uccessful numerical approximation technique for Bayesian sequential estimation with non-linear, non-Gaussian models. In our application, the fast motion of hockey players and the color model we adopt =-=[12, 13]-=- is highly non-linear and non-Gaussian. Therefore, particle filtering is the ideal model to be the basic skeleton of our tracking system. The basic Bayesian filtering is a recursive process in which e... |

342 | Rao-blackwellised particle filtering for dynamic bayesian networks
- Doucet, Freitas, et al.
- 2000
(Show Context)
Citation Context ... The key idea is to sample for the data association while the remaining tracking part is computed in an analytical way, for example, the Kalman filter. An in-depth explanation of RBPF can be found in =-=[DdFMR00]-=-. Because the Monte Carlo methods sample in the state space, the “curse of dimensionality” is the major problem if the total number of targets is large. Pérez et al. tried to solve the problem by deve... |

312 | Bramble: A Bayesian multiple-blob tracker
- Isard, MacCormick
- 2001
(Show Context)
Citation Context ...Cai, N. de Freitas, and J.J. Little target shape modelling can help resolving the likelihood modelling and data association problems during occlusions. The approach is often used within static scenes =-=[8, 9, 10]-=-. However, in our application, camera motion makes it difficult to separate target motion or perform background subtraction. Players with drastic pose changes are difficult to be captured by any expli... |

299 | A boosted particle filter: Multitarget detection and tracking
- Okuma, Taleghani, et al.
- 2004
(Show Context)
Citation Context ...ayer tracking. Particle filtering was first introduced to visual tracking by Isard and Blake in [1]. Pérez et al. [2, 3] extended the particle filter framework to track multiple targets. Okuma et al. =-=[4]-=- further extended it [3] by incorporating a boosting detector [5] into the particle filter for automatic initialization of a variable number of targets. However, as their system did not have explicit ... |

290 | Recognising panoramas
- Brown, Lowe
- 2003
(Show Context)
Citation Context ... Hij from frame j to frame i can be parameterized by the parameters of the camera in the two time steps, which is 8 in all. Details of how to form the homography with the 8 parameters can be found in =-=[BL03]-=-. With this technique, it is possible to build a whole panorama for the scene, including the audience, from a sequence of video frames. It allows many other techniques to assist in improving the track... |

275 | Human detection based on a probabilistic assembly of robust part detectors - Mikolajczyk, Schmid, et al. - 2004 |

258 | A probabilistic exclusion principle for tracking multiple objects
- MacCormick, Blake
- 1999
(Show Context)
Citation Context ...Cai, N. de Freitas, and J.J. Little target shape modelling can help resolving the likelihood modelling and data association problems during occlusions. The approach is often used within static scenes =-=[8, 9, 10]-=-. However, in our application, camera motion makes it difficult to separate target motion or perform background subtraction. Players with drastic pose changes are difficult to be captured by any expli... |

217 | and motion from image streams: a factorization method
- Tomasi, Kanade, et al.
- 1991
(Show Context)
Citation Context ...approach is to represent an object or an image 8sby a set of local features, which are unique and distinguish the object from others. The Harris corners [HS88], the Kanade-Lucas-Tomasi (KLT) features =-=[TK91]-=- and the Scale Invariant Feature Transform (SIFT) [Low04] are three typical examples. These features can be applied to specific object recognition and tracking, image matching and rectification, 3D st... |

213 |
An efficient implementation of Reid’s multiple hypothesis tracking algorithm and its evaluation for the purpose of visual tracking
- Cox, Hingorani
- 1996
(Show Context)
Citation Context ...n the assumption that uncertainties can be resolved at subsequent stages. MHT was first called Reid’s algorithm. An efficient implementation of Reid’s approach is later presented by Cox and Hingorani =-=[CH96]-=-. Multiple hypotheses are maintained at each step. New hypotheses are generated according to the previous hypotheses. Each hypothesis in the previous stage will generate one or more children. Therefor... |

211 | The unscented particle filter
- Merwe, Doucet, et al.
- 1997
(Show Context)
Citation Context ...propogated at time t by the proposal distribution to represent to make it a equallly weighted particle set {x (i) t−1 , N −1 } N i=1 to {˜x (i) t , N −1 } N i=1 p(xt|y0:t−1) and iterations go forward =-=[vdMAdFW00]-=-. . . . . . . . . 23 3.2 The final proposal is in the form of a mixture of Gaussian which combines the prior and the boosting detection [OTdF + 04]. . . . . . . 24 4.1 This shows how video frames are ... |

205 | Mean-shift blob tracking through scale space - COLLINS |

166 | Data fusion for visual tracking with particles
- Perez, Vermaak, et al.
- 2004
(Show Context)
Citation Context ...ple cues are used to resolve the ambiguity during occlusion. Wu and Huang [WH01] introduced a co-inference tracking algorithm to utilize both color and shape cues to track single target. Pérez et al. =-=[PVB04]-=- fuse color, motion and sound cues to track multiple objects with a particle filter. In our application, according to the law of inertia, it is easier to use the physical model to predict the motion o... |

155 | Tracking multiple moving targets with a mobile robot using particle filters and statistical data association - Schulz, Burgard, et al. - 2001 |

143 | Finding and tracking people from the bottom up
- Ramanan, Forsyth
- 2003
(Show Context)
Citation Context ...me geometric constraints [MSZ04, RF03]. As is mentioned in the previous section, because of the low resolution of the input video clips, tracking parts of the human body separately in the same way as =-=[RF03]-=- does is almost infeasible in this application. In the systems that detect or track people as a whole, human bodies are more or less upright and straight so that most of the bodies fit into a rectangu... |

133 | Maintaining multi-modality through mixture tracking
- Vermaak, Doucet, et al.
- 2003
(Show Context)
Citation Context ...ddress the problem of robust multi-target tracking within the application of hockey player tracking. Particle filtering was first introduced to visual tracking by Isard and Blake in [1]. Pérez et al. =-=[2, 3]-=- extended the particle filter framework to track multiple targets. Okuma et al. [4] further extended it [3] by incorporating a boosting detector [5] into the particle filter for automatic initializati... |

133 | Tracking multiple humans in complex situations
- Zhao, Nevatia
(Show Context)
Citation Context ...ve been taken to solve the occlusion problem in tracking. Kang et al. [6] tried to resolve the ambiguity of the locations of the targets by registering video frames from multiple cameras. Zhao et al. =-=[7]-=- also rectified video frames to the predefined ground plane and model the targets in the 3D space with a body shape model. A static camera was used and background subtraction was applied as well in th... |

120 |
Linear Network Optimization: Algorithms and Codes
- Bertsekas
- 1991
(Show Context)
Citation Context ... on, it was found that the auction algorithm [16] is the most efficient method so far to reach the optimal solution or sub-optimal one without any practical difference. The extended auction algorithm =-=[17]-=- is able to solve the rectangular matrix problems with the constraint that one observation can only be assigned to one target while a target can have at least one observations. However, in our applica... |

100 | Tracking multiple objects with particle filtering
- Hue, Cadre, et al.
(Show Context)
Citation Context ...ddress the problem of robust multi-target tracking within the application of hockey player tracking. Particle filtering was first introduced to visual tracking by Isard and Blake in [1]. Pérez et al. =-=[2, 3]-=- extended the particle filter framework to track multiple targets. Okuma et al. [4] further extended it [3] by incorporating a boosting detector [5] into the particle filter for automatic initializati... |

93 | A Solution to the Simultaneous Localisation and Map Building (SLAM) Problem
- Dissanayake, Newman, et al.
- 2001
(Show Context)
Citation Context ...cific object recognition and tracking, image matching and rectification, 3D structure reconstruction from 2D images for scene or motion analysis, and even Simultaneous Localization and Mapping (SLAM) =-=[DNCC01]-=- for robotics. In our work, KLT features are used for image rectification [OLL04]. As is mentioned in Section 1.3.1, the locations of the hockey players in the input video are in temporally variant im... |

88 | Learning and classification of complex dynamics
- North, Blake, et al.
- 2000
(Show Context)
Citation Context ...egression coefficients, and Zt is the stochastic disturbance which is normally a white noise process. The autoregression coefficients can either be learned or predefined in an adhoc way. North et al. =-=[NBIR00]-=- tried to determine the coefficients by learning the 30 xt r xt f ytsx k,t f Figure 4.3: All the black rectangles are the boosting detections. Black dots on the bottom edge of each rectangle represent... |

85 | Monte Carlo filtering for multitarget tracking and data association
- Vermaak, Godsill, et al.
- 2005
(Show Context)
Citation Context ...roblem by developing three algorithms: Monte Carlo Joint Probability Data Association Filter (MC-JPDAF), Sequential Sampling Particle Filter (SSPF) and Independent Partition Particle Filter (IPPF) in =-=[VGP05]-=-. The fundamental motivation underlying these three algorithms is to factorize the joint state space into individual components. However, in visual tracking systems, detections at each stage can be co... |

79 | People tracking with anonymous and id-sensors using Rao-Blackwellised particle filters - Schulz, Fox, et al. - 2003 |

59 | A co-inference approach to robust visual tracking
- Wu, Huang
- 2001
(Show Context)
Citation Context ...ckey players in our application makes the shape modelling difficult. While the mutual exclusion is difficult to achieve, multiple cues are used to resolve the ambiguity during occlusion. Wu and Huang =-=[WH01]-=- introduced a co-inference tracking algorithm to utilize both color and shape cues to track single target. Pérez et al. [PVB04] fuse color, motion and sound cues to track multiple objects with a parti... |

44 | An experimental and theoretical investigation into simultaneous localisation and map building - Dissanayake, Newman, et al. - 1999 |

37 |
Tracking appearances with occlusions
- Wu, Yu, et al.
- 2003
(Show Context)
Citation Context ...rinciple for tracking multiple objects in [MB00]. MacCormick also introduced a Bayesian multiple-blob tracker [IM01], which explicitly address the issue of mutual exclusion in the 3D space. Wu et al. =-=[WYH03]-=- explicitly model overlapping to handle occlusion. However, all the approaches in [MB00, IM01, WYH03] require explicit modelling of the object shape. As is mentioned in Chapter 1, the non-rigidity of ... |

32 |
Simultaneous estimation of segmentation and shape
- Rittscher, Tu, et al.
- 2005
(Show Context)
Citation Context ...Cai, N. de Freitas, and J.J. Little target shape modelling can help resolving the likelihood modelling and data association problems during occlusions. The approach is often used within static scenes =-=[8, 9, 10]-=-. However, in our application, camera motion makes it difficult to separate target motion or perform background subtraction. Players with drastic pose changes are difficult to be captured by any expli... |

29 | Bayesian object localisation in images
- Sullivan, Blake, et al.
(Show Context)
Citation Context ...ed by multiple targets and one target can generate multiple signals. When targets occlude each other, mutual exclusion plays a critical role for the evaluation of likelihood function. Sullivan et al. =-=[SBIM01]-=- address the global optimal solution to associate pixels to objects and background in a image with a Baysian approach. It is important for multitarget tracking because it makes the likelihood evaluati... |

27 | Real time hand tracking by combining particle filtering and mean shift
- Shan, Wei, et al.
- 2004
(Show Context)
Citation Context ...he existing tracks. Finally, the mean-shift algorithm is embedded into the particle filter framework to stabilize the trajectories of the targets for reliable motion prediction. Although similar work =-=[11]-=- has been done on combining mean-shift with particle filtering, our work is the first one that describes in detail the theoretical formulation of embedding mean-shift seamlessly into the particle filt... |

26 |
Bayesian Methods for Tracking
- Gordon
- 1993
(Show Context)
Citation Context ... sample set. As a result, the original particles that have higher weights would have more replicates while those with small weights would have much less or even no replicates. Details can be found in =-=[Gor94]-=-. The whole process of particle filtering is shown in Figure 3.1 [vdMAdFW00]. 22sFigure 3.1: The SIR process starts at time t − 1 with a set of equally weighted particles {˜x (i) t−1 , N −1 } N i=1 , ... |

25 | Rao-Blackwellized Monte Carlo data association for multiple target tracking - Särkkä, Vehtari, et al. - 2004 |

20 | time face and object tracking as a component of a perceptual user interface - Real - 1998 |

18 |
Automatic rectification of long image sequences
- Okuma, Little
(Show Context)
Citation Context ...e system changes over time with respect to the hockey rink coordinates. Therefore, target motion modelling and prediction in the image coordinates are difficult. We adopt the approach by Okuma et al. =-=[14]-=- to map the locations of the targets from the image coordinates to the standard hockey rink coordinate system which is consistent over time. Therefore, according to the physical law of inertia, the mo... |

13 | Soccer player tracking across uncalibrated camera streams
- Kang, Cohen, et al.
- 2003
(Show Context)
Citation Context ...ual occlusions between targets, it loses the identities of the targets after occlusions. On the other hand , various approaches have been taken to solve the occlusion problem in tracking. Kang et al. =-=[6]-=- tried to resolve the ambiguity of the locations of the targets by registering video frames from multiple cameras. Zhao et al. [7] also rectified video frames to the predefined ground plane and model ... |

13 | 2002), Parametric contour tracking using unscented Kalman filter, paper presented at
- Chen, Huang, et al.
(Show Context)
Citation Context ...rough the true nonlinear function to capture the true distribution. Because of its capability of handling nonlinear models and its efficiency in computation, it has been successfully used in tracking =-=[CHR02]-=-. Particle Filter The particle filter gained its popularity because of its ability to handle highly nonlinear and non-Gaussian models in Bayesian filtering with a clear and neat numerical approximatio... |

7 | Template-based action recognition: Classifying hockey players movement
- Wu
- 2005
(Show Context)
Citation Context ...ames is determined by their relative distance to the camera and the zooming effect of the camera. Therefore, it is important to adaptively change the scale through out the tracking process. Wu et al. =-=[Wu05]-=- tried to stabilize the scales of the targets through a pyramid based template matching, which is a coarse-to-fine scale search process to 1 ɛ selects the stop criteria for the mean-shift iteration. A... |

4 |
A 3D Space Formulation of a Navigation Kalman Filter for Autonomous Vehicles. The Robotics
- Kelly
- 1994
(Show Context)
Citation Context ...solutions to the Bayesian sequential estimation if both function f(·) in Equation 2.1 and h(·) in Equation 2.2 are linear and Gaussian. It is widely used in visual tracking [CH96, KCM03] and robotics =-=[Kel94]-=-. In order to handle the nonlinear and non-Gaussian situations, extensions have been made to the standard Kalman filter. The Extended Kalman Filter (EKF) [AM79] is one of the extensions which uses a f... |

4 |
Analysis of Player Actions in Selected Hockey Game Situations
- Li, Woodham
- 2005
(Show Context)
Citation Context ...ng, running, standing) rather than their translational motion and assume that the targets are stably tracked so that the targets are roughly centered within the bounding box of the tracker. Li et al. =-=[Li04]-=- aim to represent and reason about hockey behaviors for an even higher level of motion understanding. Such applications all require accurate trajectory data extracted from lower level tracking systems... |

2 | Tracking objects from multiple stationary and moving cameras
- Kang, Cohen, et al.
- 2004
(Show Context)
Citation Context ...dinates variable with respect to the coordinate system, which is the hockey rink in this case. Multiple cameras are able to solve the problem of localizing targets in the real world coordinate system =-=[KCM04]-=- so long as the relative location and configuration of different cameras are known and well synchronized. However, because the source data we use for this task is extracted from only one monocular TV ... |

1 |
The Divergence and Bhattacharyya Distance Meatures in Signal Selection
- Kailath
- 1967
(Show Context)
Citation Context ...gram. Because the purpose is to track a specific target in the scene, a reference histogram of the particular target is created at the initialization step. A metric based on Bhattacharyya coefficient =-=[Kai67]-=- is adopted to evaluate the similarity between two color histograms. There are two major weaknesses in this color-based approach. One is the loss of spatial information because the histogram approach ... |