6 Tools
There are currently three major areas of development
in computer vision algorithms. The first is low level
computer detection algorithms. Some of these include
line, shape, and object detection. The second area is
the development of geometries from smaller less complex
components. The last major area is the development
of computer vision systems based strictly on
modeling the human visual system. New computer
vision tools will be developed and enhanced in this
project. There are currently available a large number
of statistical and frequency domain based image processing
algorithms. Therefore, these traditional image
processing algorithms will be used. The development
of new tools will concentrate on neural nets and
object identification tools. One of the new tools under
development is eigan vector parameterization of
classes of images. This technique decomposes parts of
an image into areas of similarity with common objects
and then identifies the object based on the amount of similarity to a known object. This method his
considerable advantages in data compression. These
is evidence that this is how some aspects of
recognition is done. Also, I will continue my development of new neural nets to aid in computer vision
Neural nets have proven useful in locating object
image data. The use of neural net in image processing
and computer vision is not only very powerful, but
also allows the flexibility to develop systems based on
given requirements of a project and still model the
human visual system. I believe the development of
these tools will allow us to develop more intelligent
computer vision systems and understand more of the
of the human visual system.