The 10th International Conference on [629737]

The 10th International Conference on
Development and Application Systems
DAS 2010

www.dasconference.ro

Abstracts Book

May 27-29, 2010 Suceava – Romania

Organized by:

Stefan cel Mare University of Suceava
Faculty of Electrical Engineering and Computer Science
ROMANIA

Published by : Faculty of Electrical Engin eering and Computer Science

Editors:

Adrian GRAUR
Stefan Gh. PENTIUC
Cristina TURCU
Valentin POPA
C o r n e l T U R C U
Alin Dan POTORAC
Eugen COCA

ISSN 1844-5020

Copyright 2010. All rights reserve d. No part of this book may be reproduced in any form or by
any means, without written permission from the publisher.

Wording, contents and translations quality of the papers are entirely in the responsibility of the
authors.

Editura Universit ății "Ștefan cel Mare" din Suceava

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

CONTENTS

KEYNOTES
32-accept.doc
Power-Electronics Issues of Mo dern Electric Railway Systems
A. STEIMEL ……………………………………………………………………………………………………… ……………………………………………….1
das_paper_kuntman_2010.doc
New Advances and Possibilities in Active Circuit Design
H. Hakan KUNTMAN ………………………………………………………………………………………………… ……………………………………….9

SECTION A
33-accept.doc
Option to Provide the Necessary Feedback for Closed-Loop neuroStimulation
Radu BAZAVAN, Rodica STRUNGARU……………………………………………………………………………………. ……………………….19
35-accept.doc
Obstacle Avoidance Fuzzy System for Mobile Robot with IR Sensors
C. G. RUSU, I. T. BIROU………………………………………………………………………………………….. ……………………………………….25
37-accept.doc
Hazardous Events Monitoring System in a Hospital
Radu ȚIGĂNESCUL-AMARI ȚII …………………………………………………………………………………………………………………………30
40-accept.doc
Real-time Communications for Distributed Control Systems
Dan PUIU, Florin MOLDOVE ANU, Caius SULIMAN…………………………………………………………………………………………..34
41-accept.doc
Design and Comparison of Different Switched Reluctance Machines Topologies for Automotive
Applications
Claudia MARTIS, Vlad PETRUS, Adri an-Cornel POP, Johan GYSELINCK ………………………………………………………. ……40
42-accept.doc
The VSB-01 Portable System for Monitoring Environmental Conditions
Alexandru SUCIU, Vasile BU ZDUGA, Gabriela VIZITIU…………………………………………………………………… …………………46
46-accept.doc
Computer Program for Studying the Operation of Gas Turbine Plants
Pavel AT ĂNĂSOAE, Gicu OIC Ă………………………………………………………………………………………………………………………..50
47-accept.doc
Kalman Filter Based Tracking in an Video Surveillance System
Caius SULIMAN, Cristina CRU CERU, Florin MOLDOVE ANU ………………………………………………………………….. ………..54
51-accept.doc
A Fuzzy Approach Regarding the Optimization of Statistical Process Control through Shewhart Control Charts
Alexandru-Mihnea SPIRIDONIC Ă, Marius PISLARU, Romeo-Cristian CIOBANU ………………………………………………….59
57_accept.doc
Multiagent System for Robotic Vision System
Dan FLOROIAN, Florin MO LDOVEANU, Mi hai CERNAT…………………………………………………………………………………..63
58-accept.doc
The use of Fuzzy Modelling Regarding the Assurance of Environmental Protection
Alexandru-Mihnea SPIRIDON ICA, Marius PISLARU……………………………………………………………………………………………69
I

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

Kalman Filter Based Tracking in an Video
Surveillance System
Caius SULIMAN1, Cristina CRUCERU2, Florin MOLDOVEANU3
Transylvania University of Bra șov
No. 29, Eroilor Blvd., RO-69121 Braș ov
1caius.suliman@unitbv.ro, 2cri.cruceru@yahoo.com, 3moldof@unitbv.ro

Abstract — In this paper we have developed a Mat-
lab/Simulink based model for monitoring a contact in a video
surveillance sequence. For the segmentation process and corect
identification of a contact in a surveillance video, we have used
the Horn-Schunk optical flow algorithm. The position and the behavior of the correctly detected contact were monitored with
the help of the traditional Kalm an filter. After that we have
compared the results obtained from the optical flow method
with the ones obtain ed from the Kalman filter, and we show
the correct functionality of the Kalman filter based tracking. The tests were performed using video data taken with the help
of a fix camera. The tested algorithm has shown promising
results.

Index Terms — Video Surveillance System, Optical Flow,
Kalman Filtering, Image Processing, Tracking
I. INTRODUCTION
The problem of using vision to track and understand the
behavior of humans is a very important one. The main ap-
plications that it has are in the areas concerning human-ro-
bot interaction [7], robot learning, and video surveillance.
Here we try to focus our attention on video surveillance
systems. A high level of security in public places is an ex-
tremely complex challenge. A number of technologies can
be applied to various aspects of security, including biometric
systems, screening systems, and video surveillance systems. Nowadays video surveillance systems act as large-scale vid-
eo recorders, analog or dig ital. These systems serve two
main purposes: to provide a human operator with images to
detect and react to potential th reats and recording for future
investigative purposes.
From the perspective of real-time detection, it is well
known that the human’s visual attention drops below ac-ceptable levels even if that op erator is a trained one in the
task of visual monitoring. Video analysis technologies can be applied to develop smart surveillance systems that can
aid the operator in the detec tion and in the investigatory
tasks.
For surveillance applications, the tracking problem is a
fundamental component. In video surveillance one of the most used method for tracking contacts is the particle filter
[8][10][11][13]. Another well known method in the research community is the use of the traditional Kalman filter [9]. In many cases the use of this type of filter is sufficient. This is
due to the controlled indoor and outdoor environments that
are used in the studies.
Many papers in the literature detail methods that track
single persons only [6][10], but there are also many authors
that describe different methods for the detection and track-
ing of multiple persons [2][3][5 ][11]. Most of these methods involve as testing grounds indoor environments [1][3][8][13] as well as outdoor environments [2][5][8][9],
where these methods are applied to track groups.
The objective of this paper is the development of a video
surveillance system capable of tracking a person in an out-door environment. In Section II we describe the structure of
the proposed video surveillance system. In Section III we present the method used for contact detection and the me-thod used for the extraction of useful data from the video feed. Section IV describes the Kalman filter algorithm ap-
plied in our case. In Section V and VI we present the results
obtained from the Simulink model’s simulation, the conclu-sions drawn from this study and the possible future devel-opments.
II. THE SYSTEM’S STRUCTURE
In this paper we examine the feasibility of using the opti-
cal flow algorithm in conjunction with the Kalman filter algorithm [9][12] for tracking a contact in a surveillance
scene. In order to create an algo rithm that is able to track a
contact in a scene, three differ ent, large-scale task must be
accomplished (see Fig.1). First th e algorithm needs to take
an incoming surveillance video signal and segment it into a stream of frames where contac ts are distinguished from the
background of the scene. The next step is the tracking of the
contact throughout the video sequence. Finally, the resulting
track must be processed in or der to analyze the contact’s
behavior.
For the segmentation process of the incoming video sig-
nal, the optical flow algorithm developed by Horn and
Schunk was used [4]. The optical flow algorithm approxi-mates the movement of the contact in the current frame as
referenced to the previous frame. By determining the motion
of objects, one can distinguish between the contact and the background of the scene. Afte r careful tuning and process-
ing, the output of the segmentation process is passed to the
Kalman filter algorithm for further processing.

Figure 1. The structure of th e surveillance system.
The Kalman filter is a recursive, adaptive filter that oper-
ates in the state space. It is well known for its ability to track
54

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

objects in a timely and accurate manner. The tracking algo-
rithm developed in this paper is able to process one contact
at a time.
III. THE OPTICAL FLOW ANALYSIS
One of the important blocks presented in the above
scheme is the so called optical flow analysis block. The
main purpose of this block is to determine the existence of possible contacts in the incoming video signal and process them in such manner that the Kalman filter will be able to
track them with minimal error.
In Fig.2 we will present the main component of the Opti-
cal Flow Analysis block.

Figure 2. The optical flow analysis block.
In what follows we descri be the functionality for each
component block.
A. Segmentation
In our case, the term segmentation is used to describe the
process through which a video signal passes to become a
series of binary images. At th e output of this sub-block, each
of the resulting binary images will contain black and white
areas. The black areas correspond to the portion of the frame
where no motion was detected , and the white areas corre-
spond to the portion of the frame where motion was de-tected.
The surveillance system was developed in Matlab’s Simu-
link. At first the incoming video signal is coded in the RGB color space. Because we use the optical flow to detect mo-
tion, the video signal needs to be converted to the intensity color space (see Fig.3). To estimate the optical flow between
two images we use the algorithm developed by Horn and
Schunk. In our case this algor ithm is used to compute the
optical flow between the current frame and the previous one. This is one of the tunable parameters used in our experi-
ments. Another important tunable parameter used in the op-
tical flow estimation is the smoothness factor. This parame-
ter is defined as a constraint which controls how smoothly
the velocity field of the brightness pattern in images varies throughout the image. The Horn-Schunk algorithm quanti-
fies the smoothness of the velocity filed using the magnitude
of the gradient of the optical flow velocity defined as in:
2 2
⎟⎟
⎠⎞
⎜⎜
⎝⎛
∂∂+⎟
⎠⎞⎜
⎝⎛
∂∂
yu
xuand 2 2
⎟⎟
⎠⎞
⎜⎜
⎝⎛
∂∂+⎟
⎠⎞⎜
⎝⎛
∂∂
yv
xv (1)
where and u vare the velocity vectors corresponding to the
optical flow. A small value for this gradient indicates that
the vector field is very smooth; a higher one indicates the
contrary. A smooth vector field tends to zero-out regions where no motion is detected leaving only limited areas of non-zero vector fields. In the Simulink model of our sur-
veillance system, this smoothness factor is inversely pro-
portionally to the magnitude of the velocity gradients. Our experiments pointed out that the optimal value for the smoothness factor is 0.6.

Figure 3. The segmentation process.
Before the processed video signal exits the segmentation
sub-block it is compared with a certain threshold to keep only what interests us from the video feed.
B. Median Filtering
One of the biggest problems that optical flow has it’s that
it is very sensitive to changes in illumination or to the qual-ity of the video feed. This sensitivity conduces in erroneous blobs appearing in individual frames. If these blobs are
large, that means that they are approaching the average size of a real person, they can create problems for successful
morphological operations.

Figure 4. Result after the median filtering.
One of the main reason for choosing the median filter is
that most of these abnormalities , the erroneous blobs, appear
in singular frames and they do not appear again for several more frames. The median filter is used to decrease the effect
of these abnormalities while still maintaining the informa-tion of the correctly detected contacts.
In Fig.3 it can be seen that the segmented image contains
many abnormalities. After the median filtering many of these abnormalities are gone (see Fig.4).
C. Morphological Operations
The morphological operation will process the video signal
coming from the output of the median filtering sub-block in such way that all erroneous bl obs residing in the image are
eliminated and all and only the correctly detected blob is
maintained and classified as a real contact. The main mor-phological operations used by us in this study are the
ero-
sion and dilatation . Optimal erosion is achieved when the
55

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

structuring element keeps at l east the remnant of a blob for
all correct contacts. If we use a sub-optimal structuring ele-
ment for the erosion and dilatation operations, a valid con-
tact could be lost completely, or an erroneous blob could be tracked. Both these errors can produce significant barriers to
optimal dilatation and to Kalman filter tracking. Optimal
dilation is obtained when the structuring element merges all remnants of a single blob into one contact. If a sub-optimal
structuring element is used for dilation, one contact could be
viewed as multiple contacts or multiple contacts could be
viewed as one contact. After an optimal structuring element
for erosion was determined, each frame was eroded using the chosen structuring element. Determination of the opti-mal structuring element for dilation was similar to that of
erosion. Each frame was dilated with a square structuring
element.
An infinite number of possibilities exist for size and shape
of structuring elements. Depending on the data used, the size
and shape of the optimal structuring element could vary sig-
nificantly.

Figure 5. The result of morphological operations applied on the median
filtered image.
Comparing the morphological altered image to the image
resulting after the median filtering (see Fig.5) shows that the
erosion operation removed the remaining erroneous blobs
residing in the image, thus d eciding that they were not con-
tacts. Further dilation has created a solid blob out of the area where motion was detected, and this blob will be tracked as
a contact.
D. Blob Analysis
The main functionality of the blob analysis sub-block is to
determine the minimum size of a blob and the maximum
number of blobs that will be used in the Kalman tracking process and in the visualization step. By setting the minmum blob size we obtain a new level of protection against abnor-
malities by specifying a minimum size that a blob must have
in order to be correctly tracked . Thus, any blob that doesn’t
fulfill this condition will not be tracked.
The other tunable parameter of the blob analysis sub-
block, the maximum number of blobs, is used to set the number of Kalman filters to be used in the tracking process. In our case this parameter was set to 1.
IV. KALMAN FILTERING
Filtering is a very used method in engineering and em-
bedded systems. A good filtering algorithm can reduce the noise from signals while retaining the useful information. The Kalman filter is a mathematical tool that can estimate
the variables of a wide range of processes. It estimates the
states of linear systems. This type of filter works very well
in practice and that is why it is often implemented in em-bedded control system and because we need an accurate estimate of the process variables. The discrete Kalman filter
is characterized by both a process model and a measurement
equation.
The process model is character ized by the assumption that
the present state, , can be related to the past state, ,
as follows: kx1−kx
k k k kw x x+Φ=−1 (2)
where is assumed to be a discrete, white, zero-mean
process noise with known covariance matrix, ; kw
kQkΦ
represents the state transition matrix which determines the relationship between the present state and the previous one.
In our case we try to track the state of a contact based on
its last known state. Here, the state vector consists of a two-
dimensional position expressed in Cartesian coordinates, a two-dimensional velocity and a two-dimensional acceler-
tion. By considering a constant acceleration, the state tran-
stion matrix can be determined from the basic kinematic equations as follows:
2
1 1 121t a t v s sk k k k − − −+ + = (3)
t a v vk k k1 1− −+= (4)
1−=k ka a (5)
where s is defined to be the contact’s position, v is its ve-
locity, a is the contact’s acceleration and t is the sampling
period. In a matrix form, the above equations can be written
as:
⎥⎥⎥⎥⎥⎥⎥⎥
⎦⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣⎡
⎥⎥⎥⎥⎥⎥⎥⎥
⎦⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣⎡
=
⎥⎥⎥⎥⎥⎥⎥⎥
⎦⎤
⎢⎢⎢⎢⎢⎢⎢⎢
⎣⎡
−−−−−−
1 ,1 ,1 ,1 ,1 ,1 ,
,,,,,,
1 0 0 0 0 00 1 0 0 0 01 0 1 0 0 00 1 0 1 0 05 . 0 0 1 0 1 00 5 . 0 0 1 0 1
k yk xk yk xk yk x
k yk xk yk xk yk x
aavvss
aavvss
(6)
Here, the subscripts x and y refer to the direction of the
contacts position, velocity and acceleration in the two-
dimensional plane. The value of the sampling period is set to 1.
The measurement equation is defined as:
k k k kv x H z+= (8)
where represents the measurement vector, is assumed
to be a discrete, white, zero-m ean process noise with known
covariance matrix, . The matrix describes the rela-
tionship between the measurement vector, , and the state
vector, . Given the fact that the state vector is of length
six and the measurement vector is of length two, the matrix
must be of length six by two: kzkv
kRkH
kz
kx
kH
⎥⎦⎤
⎢⎣⎡=5 . 0 0 1 0 1 00 5 . 0 0 1 0 1
kH (9)
From the process model and measurement equation it re-
sults that the Kalman filter attempts to improve the prior
state estimate using the incoming measurement which has
been corrupted by noise. This improvement can be achieved
by linearly blending the prior state estimate, , with the
noisy measurement, , in: 1ˆ−kx
kz
56

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

)ˆ ( ˆ ˆ− −− + =k k k k k kx H z K x x (10)
Here means the a-priori estimate; is known as the
blending factor. The minimum mean squared error of the
estimate is obtained when the blending factor assumes the
value of the Kalman gain: −
kxˆkK
1) (− − −+ =kT
k k kT
k k k R H P H H P K (11)
where is known as the state covariance matrix. Gener-
ally the state covariance matrix is a diagonal matrix. The
state covariance matrix is dete rmined from the a-priori state
covariance matrix as follows: kP
−− =k k k k P H K I P) ( (12)
After the Kalman gain has been computed, and the state
and state error covariance matrices have been updated, the
Kalman filter makes projections for the next value of .
These projections will be used as the a-priori estimates dur-
ing processing of the next frame of data. k
k k kx xˆ ˆ1Φ =−
+ (13)
kT
k k k kQ P P+ Φ Φ =−
+1 (14)
The above equations are the projection equations for the
state estimate and for the state covariance matrix.
The main role of the Kalman filtering block is to assign a
tracking filter to each of the m easurements entering the sys-
tem from the optical flow analysis block. For an easy im-
plementation of the Kalman filter in Simulink, we wrote an
embedded Matlab function. This method is often used when the function that needs to be implemented is more easily to express in Matlab’s symbolic language than in Simulink’s
graphical language.
V. POST PROCESSING
The last block that we will discuss is the post process-
ing/video output block. This block was used to process the output from the optical flow analysis block and the output of
the Kalman filtering block.
The post processing block is composed of four video out-
put sub-blocks. The first sub-block is used only to view the original video signal.

The second sub-block shows the optical flow lines su-perimposed on the original video signal (see Fig.6a). It is
used only for scientific purposes, and that is to visualize the
direction of the optical flow lines and from there to deduce in which direction the contact moves.
The third sub-block is used to visualize the resulting sig-
nal from the optical flow analysis block and to allow the user to be sure of the correct functionality of the optical flow analysis block. This sub-block is in direct connection with
the blob analysis sub-block, block that produces the coordi-
nates for a bounding box. This bounding box is a rectangle drawn around each correctly det ected blob. The user is able
to watch in real-time which contact in the video feed is be-
ing sent to the Kalman filtering block. If rectangles are not
surrounding the correctly detected contacts in an image, this
thing means that the optical flow analysis bloc is not work-
ing properly. Fig.6b presents the resulting output of the
optical flow video viewer sub-block. We present only five
frames taken at 17.5 FPS of each other. It can be clearly
seen that a contact was dete cted and a bounding box was
correctly superimposed on the contact.
The last sub-block discussed in this section is the Kalman
filtering video viewer. This sub-block is in direct connection with the Kalman filtering bloc k. This block produces at its
output a matrix containing the position of the detected con-
tact. The output is used by the Kalman filtering video viewer
to draw markers in the video. These markers are represented
here by red circles. The center of the circle represents the coordinates of the estimated pos ition of the detected contact.
The user is able to see in real -time if the detected contact is
correctly tracked by the Kalman filter. If a marker doesn’t
follow consistently a contact we can say that the Kalman
filtering block isn’t working properly. Fig.7a presents five
frames resulting from the Kalman filtering video viewer,
taken like in the previous case , at 17.5 FPS of each other.
We can clearly see that the marker is correctly tracking the detected contact thus confirming that the Kalman filter is
working properly. The Kalman filter is even capable to track
the contact that is leaving at so me time the visual field of the
camera and then correctly rea ssign the marker to the contact
that reenters in the vi sual field (see Fig.7b).

(a)

(b)
57

10th International Conference on DEVELOPMENT AND APPLICATION SYSTEMS , Suceava, Romania, May 27-29, 2010

Figure 6. Output sample from the optical flow video viewer. (a) the op tical flow lines; (b) bounding box superimposed on the correctly de tected contact.

(a)

(b)
Figure 7. Output sample from the Kalman filtering video viewer. (a) trac king a contact that passes through the camera’s field of view; ( b) contact exiting
and reentering the camera’s field of view.
VI. CONCLUSION
There are two main factors th at affect the problem of
tracking: the accuracy to distinguish between contacts pass-
ing through the scene and the speed to process the video
feed in real-time. In this paper we have shown that with the help of the optical flow and Kalman filter algorithms it is possible to detect and track a person passing through a
scene.
The video signal used in our experiments it is provided by
a Linksys WVC200 PTZ IP video camera at a resolution of
240×320. The entire experiment was conducted using an
Intel Core2Duo T9300 computer with 4 GB of RAM.
From the optical flow analysis used in our research we
have deduced that there is an inevitable trade-off between
the accuracy and speed of proce ssing. To accurately distin-
guish a contact that passes through a scene, the computa-tional time of the optical flow algorithm must be increased. If this increase of the processing time is too large, the algo-rithm will not operate in real-time.
The Kalman filter algorithm presented in this research
was able to correctly process a contact and to correctly as-
sign a filter to the processed contact. After reviewing the results we deduced that the algorithm performed quite well showing a moderate consistency in tracking. Due with the
success with the data used in our experiments, any incon-
sistencies in the tracking proce ss can be traced back to the
fluctuations in performance of the optical flow algorithm.
Future research in the area of surveillance systems should
be focused in two directions. First, research should be made to determine an objective m easure of performance of the
optical flow algorithm and to see if other existing algorithms are better suited to accurate, r eal-time processing of a video
signal. The second research should be focused in the area of
determining contact behavior, and in areas such as merging
contacts into groups or dividing groups into separate con-tacts. For a future research, we will try to implement the pre-sented Kalman filter algorithm into a system that is capable
of tracking multiple persons.
ACKNOWLEDGMENTS
This paper is supported by the Sectoral Operational Pro-
gramme Human Resources Development (SOP HRD), fi-
nanced from the European Social Fund and by the Roma-nian Government under the contract number POSDRU/6/1.5/S/6.
REFERENCES
[1] M. A. Ali, S. Indupalli, B. Bouf ama, "Tracking Multiple People for
Video Surveillance," First Intern. Workshop on Video Processing for
Security, June 2006.
[2] B. Benfold, I. Reid, "Guiding Vi sual Surveillance by Tracking Human
Attention," Proc. of the 20th British Machine Vision Conf., Sep-
tember 2009.
[3] L.M. Fuentes, S.A. Velastin, "F rom tracking to advanced surveil-
lance," Proc. of the Intern. Conf . on Image Processing, vol. 3, pp.
121-124, September 2003.
[4] K.P. Horn, B.G. Schunck, "Deter mining optical flow," Artificial
intelligence, vol. 17, pp. 185-203, 1981.
[5] C. C. Hsieh, S. S. Hsu, "A Simp le and Fast Surveillance System for
Human Tracking and Behavior Analysis," Proc. of the 3rd Intern.
IEEE Conf. on Signal-Image Techno logies and Internet-Based Sys-
tem, pp. 812-818, December 2007.
[6] F. Jean, R. Bergevin, A.B. Albu, "Body tracking in human walk from
monocular video sequences," Proc. of the 2nd Canadian Conf. on Computer and Robot Vision, pp. 144-151, May 2005.
[7] N. Koenig, "Toward real-time huma n detection and tracking in di-
verse environments," Proc. of the 6th IEEE Intern. Conf. on Devel-opment and Learning, pp. 94-98, July 2007.
[8] S. Kong, M.K. Bhuyan, C. Sander son, B.C. Lovell, "Tracking of
Persons for Video Surveillance of Unattended Environments," Proc. of the 19th Intern. Conf. on Pattern Recognition, pp. 1-4, December 2008.
[9] W. Niu, L. Jiao, D. Han, Y. F. Wang, "Real-time multiperson tracking
in video surveillance," Proc. of the 4th Pacific Rim Conf. on Multi-media, vol. 2, pp. 1144-1148, December 2003.
[10] A.W. Senior, G. Potamianos, S. C hu, Z. Zhang, and A. Hampapur, " A
comparison of multicamera person-tracking algorithms ," Proc. IEEE
Int. Works. Visual Surveillance, May 2006.
[11] J. Wang, Y.g Yin, H. Man, "Multiple Human Tracking Using Particle Filter with Gaussian Process Dy namical Model," EURASIP Journal
on Image and Video Processing, vo l. 2008, Article ID 969456, 10
pages, 2008.
[12] G. Welch, G. Bishop, "An Introducti on to the Kalman Filter,", Tech-
nical Report: TR95-041, University of North Carolina, 2006.
[13] J. Yao, J.M. Odobez, "Multi-Camera 3d Person Tracking With Parti-cle Filter In A Surveillance Environm ent, "Proc. of the 16th European
Signal Processing Conf., August 2008.

58