ESTIMATING CROWD DENSITY WITH
MINKOWSKI FRACTAL DIMENSION
A.
N
Maranal,
L.
da
F.
Costa2,
R.
A.
Lotufo3,
S.
A.
Velastin4
1UNESP
-
Rio Claro
-
SP
-
Brazil
(nilceu@demac.igce.unesp.br)
2USP
-
S2o
Carlos
-
SP
-
Brazil
(luciano@ifsc.sc.usp.br)
3UNICAMP
-
Campinas
-
SP
-
Brazil
(lotufo@dca.fee.unicamp.br)
4KCL
-
University
of
London
-
London
-
UK
(s.velastin@kcl.ac.uk)
ABSTRACT
The estimation of the number of people in an area under
surveillance is very important for the problem
of
crowd
monitoring. When an area reaches an occupation level greater
than the projected one, people’s safety can be in danger. This
paper describes a new technique for crowd density estimation
based on Minkowski fractal dimension. Fractal dimension has
been widely used to characterize data texture in a large number
of physical and biological sciences. The results of
our
experiments show that fractal dimension can also be used to
characterize levels
of
people congestion in images of crowds.
The proposed technique is compared with a statistical and a
spectral technique, in a test study
of
nearly
300
images of a
specific area of the Liverpool Street Railway Station, London,
UK.
Results obtained in this test study are presented.
1. INTRODUCTION
The management and control
of
crowds
is
a crucial problem for
human life and safety [1,4,12]. One important aspect of the
correct management and control
of
crowd is the real-time
monitoring of crowds, which
is
usually carried out by means of
extensive closed-circuit television systems and human observers.
In order to prevent accidents, mainly in routine monitoring, such
as
those carried out in airports, train and subway stations, the
automation of this job is a necessity.
This paper presents a technique for automatic crowd density
estimation based on Minkowski fractal dimension of the image
of the area under monitoring. Fractal geometry provides an
elegant framework that can be taken as basis for a theoretical
analysis and characterization of complex curves exhibiting some
kind of intrinsic self-similarity at different magnification scales
[81.
0-7803-5041-3199
$10.00
O
1999
IEEE
3521
The self-similarity dimension (also known
as
box-counting
or
Minkowski dimension) provides a measure of the degree of
‘space-filling’ exhibited by
a
particular fractal curve [15], and
has been increasingly applied
as
a means of characterizing data
texture and shape in a large number of physical and biological
sciences [2,5,11,13].
This paper is organized as follows. In Section 2 we present the
technique for crowd density estimation based on Minkowski
fractal dimension. In Section 3 we discuss how the fractal
dimensions of crowd image can be estimated by means of
Minkowski sausages. In Section
4
we present the results
obtained by the new technique and compare them with results
obtained by two other techniques: one statistical and one
spectral. Finally, in Section 5, we summarize
our
work and
findings.
2. CROWD DENSITY ESTIMATION
Figure
1
shows a diagram
of
the proposed technique for crowd
density estimation by using the Minkowski fractal dimension.
First, it is carried out edge detection in the input image and a
binary image is generated. Next,
n
dilations are computed from
the binary image with structuring elements
of
different sizes,
ranging from
I
to
n.
From these dilations, it is estimated the
fractal dimension of the input image
(see
details in Section
3).
Finally, the fractal dimension is used
as
the feature to classify the
image as belonging to one of the five following classes: VL
-
very low density; L
-
low density; M
-
moderate density; H
-
high density; or VH
-
very high density.