Information Technology Reference
In-Depth Information
Football Players Classification in a Multi-camera
Environment
Pier Luigi Mazzeo, Paolo Spagnolo, Marco Leo, and Tiziana D'Orazio
Istituto di Studi sui Sistemi Intelligenti per l'Automazione, C.N.R.
Via G. Amendola 122/D 70126 Bari, Italy
{mazzeo,dorazio,leo,spagnolo}@ba.issia.cnr.it
http://www.issia.cnr.it/
Abstract. In order to perform automatic analysis of sport videos ac-
quired from a multi-sensing environment, it is fundamental to face the
problem of automatic football team discrimination. A correct assignment
of each player to the relative team is a preliminary task that together
with player detection and tracking algorithms can strongly affect any
high level semantic analysis. Supervised approaches for object classifi-
cation, require the construction of ad hoc models before the processing
and also a manual selection of different player patches belonging to the
team classes. The idea of this paper is to collect the players patches com-
ing from six different cameras, and after a pre-processing step based on
CBTF (Cumulative Brightness Transfer Function) studying and compar-
ing different unsupervised method for classification. The pre-processing
step based on CBTF has been implemented in order to mitigate differ-
ence in appearance between images acquired by different cameras. We
tested three different unsupervised classification algorithms (MBSAS - a
sequential clustering algorithm; BCLS - a competitive one; and k-means
- a hard-clustering algorithm) on the transformed patches. Results ob-
tained by comparing different set of features with different classifiers are
proposed. Experimental results have been carried out on different real
matches of the Italian Serie A.
1
Introduction
In last years sport applications of computer vision are increasing in many contexts:
in particular, many works focus on football applications, since it is one among the
most popular team sports around the world, and it has a large audience in all the
television programs. The research activities in sports video have focused mainly
on semantic annotation [1], event detection [14] and summarization [3]. The high
level applications above mentioned are based on structural low level procedures:
the player segmentation [4], tracking [10] and their classification [6].
In this work we focus our attention mostly on the last aspect of image analysis:
the automatic classification of players according to their team membership in a
multi-camera context. Automatic team discrimination is very important because
it allows to both reduce the interaction of human people and make the whole
 
Search WWH ::




Custom Search