A.M. Baumberg, "Learning Deformable Models for Tracking Human Motion", PhD thesis, The University of Leeds, School of Computer Studies, UK, October 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....We show image filtering methods which can enhance edge detection, thereby improving the robustness of image analysis software for people tracking. 1 People Tracking The tracking system used for our experiments is an extension of the Leeds People Tracker which was developed by Baumberg and Hogg [1]. We have ported the tracker from an sgi TM platform to a PC running GNU Linux to make economic system integration feasible. The tracker uses an active shape model [1] for the contour of a person in the image. A space of suitable models is learnt in a training stage using a set of video images ....

....tracking system used for our experiments is an extension of the Leeds People Tracker which was developed by Baumberg and Hogg [1] We have ported the tracker from an sgi TM platform to a PC running GNU Linux to make economic system integration feasible. The tracker uses an active shape model [1] for the contour of a person in the image. A space of suitable models is learnt in a training stage using a set of video images containing walking pedestrians. Detected person outline shapes are represented using cubic Bsplines, giving a large set of points in a highdimensional parameter space. ....

[Article contains additional citation context not shown here]

Adam M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, October 1995.

....can be classified into three main categories of increasing complexity: 1. Methods using region or blob based tracking, sometimes with additional classification schemes based on colour, texture or other local image properties [2 4, 6, 7] 2. Methods using 2D appearance models of human beings [1, 5]. 3. Methods using full 3D modelling of human beings [8, 9] Fig. 1. View from Surveillance Camera with Tracked Outline Shapes The more detailed the models for people detection and tracking, the better the system can handle the particular situations for which it is trained. However, systems in ....

....camera scene setup [8] rarely available in a visual surveillance environment. This is why most of the methods used for visual surveillance fall into categories 1 or 2. The people tracker developed in this work is in category 2. Our people tracker uses a Region Tracker, and an Active Shape Model [5] to model the 2D outline of a person in the image. Figure 1 shows a number of people tracked by the Active Shape Tracker. The People Tracker has been modified over time to increase tracking robustness and adapted for use in an integrated visual surveillance system. We ported the tracker from an ....

[Article contains additional citation context not shown here]

A. M. Baumberg, Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, October 1995.

....Then, semantic object segmentation can be achieved by change detection and or motion segmentation [1] Other methods impose constraints on the shape of the tracked object by using 2D [2] 3] or 3D shape [4] 5] models. Some algorithms acquire the 2D shape space information through training [6] [7], and use projections onto the shape space to estimate the most likely object boundary at a certain frame. The CONDENSATION algorithm [8] which is a state space sampling approach, needs the shape space to be known beforehand. Another ap plication of learned motion models for tracking is ....

.... [2] state space search model, defines testing objects initial model Comaniciu Color Initial region No shape model Histogram matching Track non rigid [11] histogram marked (elliptical region) using mean shift objects (regions) Baumberg B spfines, Learn shape Eigenshapes Kalman Track human [7] uniform motion space automatically filter motion W 4 Gray scale Learn background, Silhouettes Shape Surveillance [14] images Learn projection of people analysis and of people templates tracking Proposed Performance Initial boundary None Feature tracking, Rigid : non rigid Method feedback ....

A.M. Baumberg, Learning Deformable Models for Tracking Human Motion, Ph.D. thesis, The University of Leeds, School of Computer Studies, October 1995.

....A common method is the use of a Kalman filter [11] that takes into account the position and the velocity of each object. Also, the appearance of the targets has been used. For example the Condensation tracking algorithm, proposed by Isari and A. Blake [12] 13] uses Point Distribution Models (PDMs) [14] that describes the shape of the target. Brock Gunn and Ellis propose colour appearance models to disambiguite target identities after static and dynamic occlusions [15] 1.3 Multiple Camera Surveillance Systems Even in small shops, more than one camera are used for security purposes. Although ....

A. M. Baumberg, Learning deformable models for tracking human motion. PhD thesis, University of Leeds, UK, 1995.

....Figure 2 shows an example of the visualised output from the People Tracker. 3.1. Brief History The original People Tracker was written at the University of Leeds in 1993 1995 using C running under IRIX on an sgi. It was a research and development system and a proof of concept for a PhD thesis [2]. The main focus during development was on functionality and experimental features which represented the state of the art in people tracking at that time. No software design technique was employed during the code development. The only documentation generated was a short manual on how to write a ....

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, October 1995.

....in Video Understanding C. Chang University of Illinois at Urbana Champaign Urbana, IL, 61801 USA I Introduction The detection of moving objects is an important problem in image sequence analysis. Previous approaches to this problem can be divided into the following two classes: 2D algorithms [2], 5] 6] 18] 28] and 3D algorithms. The variety of approaches is wide: representative examples can integrate learning [4] work on compressed media [8] and even detect semantic objects [1] 7] 26] Furthermore, probabilistic learning approaches appear to be promising [6] 18] which ....

....and fits it with a set of spatio temporal snakes. The snakes can be used to find the bounding contours of the walker. The contours vary over time in a manner characteristic of each walker and can be used to control a simple 5 stick model (two sticks per leg, and another for the torso) In [2], a system is presented, which addresses the analysis and automatic interpretation of images containing moving non rigid objects, such as walking people. It can be automatically trained on video sequences to learn constraints on the apparent shape and motion of a particular non rigid object in ....

A.M. Baumberg, "Learning Deformable Models for Tracking Human Motion", PhD thesis, The University of Leeds, School of Computer Studies, UK, October 1995

....which makes their movements hard to predict. Figure 1: Example footage of soccer players. 1 2 Background Many different trackers with various purposes have been developed in recent years. One of the first of these, designed for individual pedestrian surveillance, was the Leeds People Tracker [1] which employs contour tracking, active shape models, and Kalman filtering to track multiple people from a single camera. Some systems allow a determination of the body pose, and real time tracking of head and hands, such as Pfinder [16] Pfinder is a person finder which uses a multi class ....

.... image segmentation from video, using background subtraction, adaptive background subtraction [10] and colour space models [14] Maintaining a temporal background model and performing background subtraction has been shown to be a fast and efficient method of extracting moving objects from a scene [1]. This performs best in relatively empty scenes through which objects are moving, but sporting activities do not fit into this category. Sports players are always on the pitch, and often (particularly in sports like netball) have a tactical position in which they stand for short lengths of time, ....

[Article contains additional citation context not shown here]

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, 1995.

....a background image, which compensates for the particular idiosyncrasies inherent to the problem. A multi faceted approach to tracking has been demonstrated, and preliminary examination shows the approach to be robust. 1 Introduction Automated tracking via passive observation is well developed [2, 7], but has received little application to the animal domain, although analysis of snapshot behaviour has demonstrated the potential to automate some farming tasks [9] Many applications exist, however, in which the groups of animals to be studied are active, sometimes for protracted periods. One ....

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, 1995.

....We show image ltering methods which can enhance edge detection, thereby improving the robustness of image analysis software for people tracking. 1 People Tracking The tracking system used for our experiments is an extension of the Leeds People Tracker which was developed by Baumberg and Hogg [1]. We have ported the tracker from an sgi TM platform to a PC running GNU Linux to make economic system integration feasible. The tracker uses an active shape model [1] for the contour of a person in the image. A space of suitable models is learnt in a training stage using a set of video images ....

....tracking system used for our experiments is an extension of the Leeds People Tracker which was developed by Baumberg and Hogg [1] We have ported the tracker from an sgi TM platform to a PC running GNU Linux to make economic system integration feasible. The tracker uses an active shape model [1] for the contour of a person in the image. A space of suitable models is learnt in a training stage using a set of video images containing walking pedestrians. Detected person outline shapes are represented using cubic Bsplines, giving a large set of points in a highdimensional parameter space. ....

[Article contains additional citation context not shown here]

Adam M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, October 1995.

....to locate the face. The iterative search procedure returns a set of shape parameters describing the best match found of the model to the data. We can also extract the shape free grey level parameters from the extracted shape, and thence calculate the combined appearance model parameters. Baumberg [1] has described a Kalman filter framework used as a optimal recursive estimator of shape from sequences using an Active Shape Model. In order to improve tracking robustness, we propose a similar scheme, based on the decoupling of identity variation from residual variation. The combined model ....

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, University of Leeds, 1995.

....been constructed based on some prior knowledge of the underlying motion or been learned automatically from the empirical regularities in observations of its many movements. In this paper we concern ourselves with the second task, assuming that the first task has been properly completed, see e.g.[1], which makes available a sequence of state vectors for us to carry on tracking. In recent years there has been substantial research towards the tracking of motions of various kinds for a variety of objects or visual curves, either rigid or non rigid [2] 3] 4] though the efforts are largely ....

....3 Neural networks for motion analysis The use of state vector sequence to describe the walking trajectory of a pedestrian makes Kalman filters an immediate choice for modelling the human motion. In fact, detailed studies have been carried out and a Kalman filter based video rate tracking system [1] has achieved some impressive results, though the system adopted several simplified conditions and linearity assumptions, such as the requirement of constant motion, uniform sampling rate, and that each shape parameter is treated (filtered) independently of the others by using a one dimensional ....

Baumberg A., Learning Deformable Models for Tracking Human Motion, PhD Thesis, School of Computer Studies, University of Leeds, October 1995.

....example. Typically, a PDM is a statistical model of a set of (2D or 3D) landmark points where each point corresponds to a particular feature on the object. The landmark points for the PDM are based on a statistical analysis of the point coordinates over a training set. In a related approach, Baumberg Hogg (1995) describe a method which tackles the problem of modelling continuous deformable contours using a spline shape representation which provides a more efficient method for calculating a statistical shape model for continuous curves rather than using a dense set of sampled boundary points. For tracking ....

....close to be distinguished, a single region will be obtained which represents several scene objects. This technique is also highly susceptible to changing lighting conditions, for example a cloud passing in front of the sun, although with more gradual changes an adaptive background can be applied. Baumberg (1995) uses this technique as a first step in his model generation process. This adaptive background technique is the method used throughout this thesis (for more details see chapters 3 and 5) It may be possible to improve the results detailed within this thesis by using a more sophisticated object ....

Baumberg, A. M. (1995), Learning Deformable Models for Tracking Human Motion, PhD thesis, University of Leeds.

....required for the particular individual. 5.2 Tracking Face Sequences In each frame of an image sequence, an Active Appearance Model can be used to locate the face. The iterative search procedure returns a set of parameters describing the best match found of the model to the data. Baumberg [1] and Rowe et al. 16] has described a Kalman filter framework used as a optimal recursive estimator of shape from sequences using an Active Shape Model. In order to improve tracking robustness, we propose a similar scheme, but using the full Appearance Model, and based on the decoupling of ....

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, University of Leeds, 1995.

....to locate the face. The iterative search procedure returns a set of shape parameters describing the best match found of the model to the data. We can also extract the shape free grey level parameters from the extracted shape, and thence calculate the combined appearance model parameters. Baumberg [1] has described a Kalman filter framework used as a optimal recursive estimator of shape from sequences using an Active Shape Model. In order to improve tracking robustness, we propose a similar scheme, based on the decoupling of identity variation from residual variation. The combined model ....

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, University of Leeds, 1995.

No context found.

A.M. Baumberg, "Learning Deformable Models for Tracking Human Motion", PhD thesis, The University of Leeds, School of Computer Studies, UK, October 1995.

No context found.

Baumberg AM. Learning deformable models for tracking human motion. PhD Thesis, School of Computer Studies, University of Leeds: Leeds, U.K., October 1995; xiv+138 pp. ftp://ftp.comp.leeds.ac.uk/comp/doc/theses/baumberg.ps.gz.

No context found.

A. M. Baumberg. Learning Deformable Models for Tracking Human Motion. PhD thesis, School of Computer Studies, University of Leeds, 1995.

No context found.

A.M. Baumberg, "Learning Deformable Models for Tracking Human Motion," PhD thesis, The University of Leeds, School of Computer Studies, UK, October 1995.

No context found.

A. M. Baumberg, "Learning deformable models for tracking human motion, " Ph.D. dissertation, School of Comput. Studies, Univ. of Leeds, Leeds, U.K., Oct. 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC