ORIGINAL ARTICLE
K. Morioka (*)
Meiji University, 1-1-1 Higashimita, Tama-ku, Kawasaki, Kanagawa
214-8571, Japan
Tel. +81-44-934-7304; Fax +81-44-934-7304
e-mail: {morioka,leejooho,h.hashimoto}@ieee.org
J.-H. Lee
Ritsumeikan University, Shiga, Japan
Y. Kuroda · H. Hashimoto
IIS, University of Tokyo, Tokyo, Japan
Artif Life Robotics (2007) 11:204–210 © ISAROB 2007
DOI 10.1007/s10015-007-0429-9
Kazuyuki Morioka · Joo-Ho Lee · Yoichi Kuroda
Hideki Hashimoto
Hybrid tracking based on color histogram for intelligent space
in the environment, such as in a house, factory, hospital,
etc.
We have proposed the concept of Intelligent Space
(iSpace)
5
in order to achieve human-centered services by
accelerating both physical and psychological interactions
between humans and their environment. The iSpace is con-
structed with CCD-camera-based intelligent sensors, which
include a processing and networking capacity. These intel-
ligent sensors are called Distributed Intelligent Network
Devices (DINDs). Positional estimations of humans and
mobile robots based on the tracking of a color marker,
6
human behavior recognition,
7
and control of mobile robots
which supporting humans
8,9
has been achieved in the
iSpace.
The most important task of the sensing infrastructure in
the iSpace is object tracking and the localization of moving
objects. The object tracking system in the iSpace can be
considered as a multi-camera multi-object tracking system.
There are two major problems in the multiple-camera
multiple-object tracking system. The first is the traditional
correspondence problem from frame to frame over time in
the image sequence. The second is the object correspon-
dence problem among different cameras in order to achieve
seamless tracking. This article focuses on the first
problem.
In the research field of object tracking, one challenging
task is how to deal with the rapid movements of the object.
A method using partial background information which does
not use all the image information, such as background sub-
traction, is presented. The benefit of using background-
weighted histograms is described in the article by Comaniciu
et al.
10
In this method, a background area which is three
times the size of the target area is extracted in every frame.
Using two histograms representing the extended back-
ground area, the features of the target object are increased
and the tracking capability is improved. In the article by
Nummiaro et al.,
11
a robust object tracker for a fast-moving
object is proposed. In this method, a hundred samples
of partial background areas are extracted in advance,
and object tracking is implemented using a particle filter.
Trackers based on background information are useful for
Abstract The vision sensor network is expected to achieve
a contact-free wide-area location system without any addi-
tional burden on users in intelligent environments. In this
article, a tracking algorithm for a location system in an
intelligent environment is described. A modified color
tracker based on a Kalman filter and a mean shift procedure
is proposed in order to improve the robustness for occlusion
and rapid movement. To handle the sudden change in
object movement, we propose a hybrid tracking algorithm,
including an adaptive feedback loop, based on the statistics
of color histogram models after the mean-shift process.
Experimental results showed that the proposed method
achieves more robust tracking of multiple objects than the
conventional method.
Key words Intelligent space · Color histogram · Kalman
filter · Hybrid tracking
1 Introduction
The research field on intelligent environments including
many ubiquitous devices such as computers and sensors, has
recently been expanding.
1,2
Some intelligent environments
utilize a distributed vision sensor network as a sensing infra-
structure.
3,4
Vision sensor networks achieve a contact-free
wide-area location system without any additional burden to
users. It also has the capacity to obtain further information
depending on the image processing. Distributed vision
sensor networks offer promising prospects as the infrastruc-
ture for robots which are required to coexist with humans
Received: December 15, 2006 / Accepted: April 16, 2007