1. Hand Gesture Recognition
KDU International Research Symposium - 2013
Maj KVPDammika, Shehan Fenando, O/C AKG Santhirathne, O/C PVVL Vitharana, O/C UJC Kumara
Department of Electrical and Electronic, Kotelawala Defence University, Kandawala Estate,
Ratmalana, Sri Lanka
gayan.shanthirathne@gmail.com>, <vibuthiya@gmail.com>, <ujchandanakumara@gmail.com>,
<kvpdhammika@gmail.com>,,<shehan117@gmail.com>
Abstract
Vision based hand gesture recognition system in practical version is a vast subject area with very
large application environment. We have seen this kind of application in gaming industries,
surveillance systems, communication systems and in military systems too. We have concentrated
much more in the communication systems and for us it was to build up a system where we could
detect the hand and get the output of how many finger are shown. This was a very hard process
because in many of other applications mentioned above has to use a special tool (a gloves) or a
pointed aid in the hand. But what we have archived in developing the process is that we can use
our bare hand for detecting purpose. This is a vital point because this suits very much in the
practical environment that we live on every day.
We used open CV and visual basic environment in order to produce the outcome. We used a
normal 1.3 webcam of laptop and we run this software in a windows operation system. Thus we
have come across on many limitations such as illumination variation and mobility of the system
ect. The count of the fingers will be shown in a LED screen where a signal comes out through
the several port to a arduino mega micro controller board.
In this we have use many predefined algorithms and method. For background subtraction we
have used code book method. For edge detection we have used canny edge detection method
and Douglas Pencker system Sklancsky algorithm used in this project also.
The prototype that we made is in good working condition but affecting few drawbacks.
Illumination variations such as shadows are not totally compatible with the software. Thus we
need to create a according environment for detection. Overcoming these drawbars would lead a
useful application for the people who are using sing language and this will be created as a mobile
application for those people