CSCI 5980-3
Color and Image Science
Image science is the study of how to capture, store,
process, transmit, and display pictures. The theory involved has been
developed in the fields of photography, printing, and television over
the past 100 years. Up until the last decade, these techniques were
only important to the relatively small number of scientists and engineers
who developed new analog imaging technologies. However, within the
last five years, the addition of digital imaging hardware to computer
workstations has made the study of image science a relevant topic
for computer scientists. User interfaces and multimedia applications
developed by software engineers now commonly include both high resolution
24 bit color images and animated sequences stored in a compressed
format. Hardcopies of black and white images are routinely made on
laser printers and high fidelity color copies are now possible using
ink jet technology. Even live television transmission over the internet
has become commonplace as network bandwidth has increased and CCD
cameras are included as workstation peripherals.
This course provides computer scientists with an
image science background that they can apply to their work. The
theory that underlies complete imaging systems (including acquisition,
coding, transmission, and display) are covered. Image compression
algorithms and image storage formats that are used to decrease the
memory necessary to represent a picture are discussed. Halftoning
and dithering algorithms that make it possible to display an image
as a pattern of dots are described. Techniques for sampling and
interpolating pictures are introduced. These interpolation techniques
include warping and morphing algorithms that are used in the movies
and in commercials to transform one object (or actor's face) into
another. Image compositing methods that are employed in the movie
industry to combine elements and create complicated special effects
are also mentioned. The new image based rendering approach to realistic
image synthesis is also overviewed. Throughout the course, special
attention is given to color and its accurate reproduction in images.
This is a lecture format course with programming
assignments given to reinforce the lecture material. A series of
video tape demonstrations developed at MIT are used in the lectures
to illustrate a number of the concepts presented. Programming assignments
may include the implementation of a warping, a morphing, an image
compression, a halftoning, and an image based rendering algorithm.
Some background in Fourier transforms and digital signal processing
are helpful but not essential. The course should be accessible to
all graduate and undergraduate students regardless of their computer
graphics background.
|
