Вы находитесь на странице: 1из 4

FACE TRACKING MOUSE - Head Tracking Driven Virtual Computer Mouse

Head Tracking technology is an alternative to the mouse that allows the person control over
the movement of the cursor by using only the movement of his head.

Abstract

We present a simple prototype system for real time tracking of a human head. This
system uses a simple yet a effective Face tracking algorithm. The general
requirements of a real time tracking algorithm – it should be computationally
inexpensive, should possess the ability to perform in different environments and
should be able to start and initialize itself with minimum knowledge about the
environment, are well addressed by the elliptical head tracking algorithm.

BACKGROUND
People who are quadriplegic and somewhat physically disabled, for example
from cerebral palsy or traumatic brain injury or stroke, have limited motions
they can make voluntarily. Some people can move their heads. Some can blink
or wink voluntarily. Some can move the eyes or tongue. Family, friends, and
other care providers usually detect these motions visually.
Many computer access methods have been developed to help people who are
quadriplegic and nonverbal: external switches, devices to detect small muscle
movements or eye blinks, head pointers, infrared or near infrared camera based
systems to detect eye movements, electrode based systems to measure the
angle of the eye in the head, even systems to detect features in EEG. These
have helped many people access the computer and have made tremendous
improvements in their lives. Still, there are many people with no reliable means
to access the computer. We are interested in developing computer vision
systems that work under normal lighting to provide computer access to people
who are quadriplegic and physically disabled.

STATEMENT OF THE PROBLEM


Develop a system that uses a camera to visually track a feature on a person’s
face, for example the tip of the nose, and use the movement of the tracked
feature to directly control the mouse pointer on a computer.
Existing System:

In the modern world, computer use has become essential for many everyday tasks
such as electronic communications, information gathering, and recreational
activities. The current computer interface set up of a mouse and keyboard requires
the user to have full use of his or her hands. Unfortunately, many people do not
have sufficient use of their hands due to injury or illness and are thus unable to use
a computer using traditional hardware

Some alternative interfaces have been developed using electroencephalograms


(EEGs) and eye motion, however these require a great deal of expensive hardware,
require significant processing time, and only give the user limited control. More
recently, development has focused on systems that monitor head motion either
electromechanically or optically. These systems can provide faster speeds and more
control, but they are often very expensive and difficult or awkward to use.
Additionally, many of these systems utilize mouth controls to some degree, which
can by hygienically troublesome.

Solution:

The project presents a simple and effective low cost optical system for
implementing mouse operations using processed head motion. The system consists
of several basic, off the shelf components including a webcam, a headset and a
computer. Images from the webcam are analyzed using a combination of software
in order to determine the position of the user’s head. This head position data is then
transformed using a non-linear transformation into a corresponding screen position
that is used to control the mouse pointer. Clicking operations can be accomplished
either using the voice commands or the still Face position.
System Architecture:
The overall system is as depicted in the block diagram given in figure 1. The image
acquisition module grabs frames from the video device which are fed into the
tracking module. Now the tracking module checks if the grabbed frame is the first
frame. A global search is performed for the first frame and the likelihood is
computed. This way the tracking module automatically initializes the tracker by
performing this global search on the first frame thus eliminating the need for any
explicit knowledge of the environment before hand. For all the subsequent frames,
the location of head is found by performing a local search in the search range.
Finally the object which is tracked in real time is displayed on the Computer screen.

Вам также может понравиться