By Soham Chakraborty,

1

• By Soham Chakraborty,
• Joey Huntley,
• and Susan Xie

2
The Project

• The premise was to use SI resources to locate
  comets in images taken from SOHO.
• SOHO is a satellite that takes pictures of the
  sun. Noise, stars, and comets are visible in the
  background, but are not distinguishable from one
  another.
• All code was to be written in MATLAB, a numerical
  computing environment and programming language.

3
Original Image
4
The Process

• First, the images were loaded into MATLAB.
• Next, the images were processed to show only
  white dots representing comets, noise, and stars
  against a black background.
• Then, we found the coordinates of all the white
  dots in each image matrix, which consisted only
  of 0s for black pixels and 1s for white pixels.

5
Processed Image
6
The Process (2)

• Next, we found the differences in time between
  subsequent images. These times varied, but were
  usually 20-40 minutes.
• Since comets move at a constant speed, around 1
  pixel per minute in our images, we were able to
  eliminate dots that didnt fall in the range. We
  tracked dot movements and considered paths over
  groups of 4 images.

7
The Process (3)

• We found all the possible paths constrained by
  the distance requirements.
• Next, we considered the angle over subsequent
  images. Comets will move in essentially straight
  lines, although some leniency was given. Paths
  that werent relatively straight were filtered
  out.
• Since SOHO has a circular orbit, the dots that
  are stars will appear to move horizontally. All
  horizontally moving objects were removed.

8
The Process (4)

• The final consideration was brightness. Fake
  comet paths might be found between random noise
  dots. Since noise varies in intensity, and
  comets do not, we eliminated all dots that varied
  significantly in brightness.
• MATLAB functions were created to perform each
  task. These were then packaged together with a
  master function and a program loop that only
  allowed a hundred images at a time to be loaded
  and processed.

9
The Results

• We ran the final function on a set of images
  known to contain comets. While it was not very
  efficient and had a long run time, our program
  was successful in isolating and tracing the
  comets.
• Thanks again to our mighty project leader, Alan
  Chalker, the master of MATLAB, for all his help.

10
THE VISUALIZATION

• Using MATLAB, a series of animation frames were
  created by using program loops to essentially
  take a shot of the images in each step of the
  visualization.
• This resulted in an animation with over 200
  frames that displays the comets as they move
  across the image groups and green lines that
  trace their paths.

11
(No Transcript)