|
Artificial eyesight
Forehead
Retina System (FRS) is a high-tech solution to what was formerly an insoluble
problem. It literally enables the blind to see. The goal is to improve
the quality of life for visually handicapped people.
The prototype is being developed with a form-factor
that resembles sunglasses. The system converts visual information into
tactile sensations using a small camera and 512 electrodes embedded inside
a headband and mounted on the forehead.
According to a WHO report in 2003, approximately
45 million people are completely blind and about 135 million have low
vision. FRS will give them a 2D representation of an object. The image
is captured by the camera then converted to tactile sensations by electrical
stimuli. An appropriate training module is necessary for using this device
successfully.
Several technologies are used for FRS.
Nerve
selective stimulation. Says Dr Hiroyuki Kajimoto, Research faculty
at the University of Electro-communications in Japan, “There are several
mechano-receptors in our skin. These are a kind of sensor that is activated
by mechanical deformation of the skin. This activity is transmitted to
our brain through nerve fibres, generating the sense of touch. Each type
of mechano-receptor is responsible for a different type of mechanical
deformation.” Independently stimulating each type of mechano-receptor’s
nerves in our skin, complex tactile sensations can be reconstructed by
combining them. This process is similar to the visual colour construction
method using RGB. It is called the ‘Tactile Primary Colour Approach.’
By appropriately designing electrical potential distribution using multiple
surface electrodes, each type of receptor can be selectively stimulated.
In FRS, the Meissner corpuscle, which responds to low frequency vibration
(of about 30 Hz), is selectively stimulated; this gives a vibratory sensation
to the person wearing the instrument. “Our electrical stimulation mainly
stimulates the Meissner corpuscle, therefore it induces a vibratory sensation
which is quite easy to perceive,” adds Kajimoto.
High speed switching. In FRS, unlike conventional
co-axial electrodes, a matrix of electrodes is used. Each electrode alternates
between anode and ground to form a virtual co-axial electrode, enabling
denser alignment.
Forehead stimulation. Sensory substitution
through electrical stimulation is old, but using the forehead as a stimulation
area is a new approach. However, forehead stimulation is quite reasonable.
It is easy to put on and take off, while coordinate system transformation
in our brain is easier than would be the case with other parts of the
body.
Image processing. An image is captured by
the camera and then converted to tactile information through two processes.
Firstly, to enhance the edges, spatial outline extraction is performed.
Then to enhance time-varying information, temporal band-pass filtering
is done. These are actually what the retina does. FRS imitates the pre-processing
done by the real visual system to facilitate image recognition.
A CCD camera attached to a pair of sunglasses captures
the view in front of the subject. After extracting the edges, the data
is converted to a tactile stimulation pattern and transmitted to the driver
circuit via a standard serial port. 512 electrodes are driven sequentially
to create the tactile pattern. The entire process is triggered by the
image capture event, which occurs every 33 ms (30 fps).
The basic electrical stimulation technology is inherited
from ‘Smart-Touch’ [Kajimoto et al. 2003] which is a visual-to-tactile
conversion system for the skin on the finger. Edge extraction is done
using luminance information followed by extraction of specific colours
using a colour key. Extraction of the outline edge is done using an ordinary
Laplacian of Gaussian (LOG) filter. After that, the image is scaled down
to a resolution of 32×16, and using threshold a black-and-white binary
pattern is obtained.
Research on FRS is being conducted by Tachi Laboratory
in the University of Tokyo and EyePlusPlus Inc. According
to them, FRS should be widely available by 2007.
— Kushal Shah
|