![[Challenge Logo]](/98-logo.gif){kind=logo}
Team Number: 047
School Name: Los Alamos High School
Area of Science: Physics - Signal and Image Processing
Project Title: 3D Image Reconstruction from X-Ray Scans
Challenge Team Interim Report
|
Team Number: 047 School Name: Los Alamos High School Area of Science: Physics - Signal and Image Processing Project Title: 3D Image Reconstruction from X-Ray Scans |
|
|
|
|
Project Summary:
Our project is to write a program which reconstructs a three
dimensional (3D) image of an object from x-ray or similar data. This
program will be parallelized to run on a multi processor machine. The
output willl be displayed using IDL, an interactive graphics language. We
are receiving the data for this project from ESA-MT at LANL, the
nondestructive testing team. They have a single processor version of the
program. Ther are providing us with x-ray data sets. It is also possible
that we will receive data from a neutron scanner. We are also using a IDL
license donated by Research Sustems Inc.
There are three distinct steps to this project. Each one is a complete project in and of itself. If we finish one we will go on to the next. If the steps prove extremely difficult we will be able to stop after step one. The first project reconstructs a 3D image from a parrallel beam. The next second step is to reconstruct a 3D image from a fan beam. The final step is to reconstruct a 3D image from a cone beam.
Each of these projects is comprised of several different parts suitable for different individuals on the team. One part is a program which calculates one slice of an image. Next a multi processor program which assembles the slices into a 3D image. Lastly the data is transferred to an IDL for 3D viewing.
Final Goal:
We expect to finish with one or more programs to take a variable
number of x-ray projections and reconstruct a 3D image of the original
object. This program will also take a simulated object create the
projections, and recreate the image. This demonstrates the error inherent
in the process. The programs, with the exception of IDL, will run on
multiple processors. We will also benchmark this program on a specific
object and compare it to the existing single processor program at LANL.
We expect to find that our multi processor program runs faster.
Current Progress:
We are currently writing the three different parts of the parallel
beam program. The coding of the parallel program will start when we get a
account on a machine with a large number of processors. We are currently
writing the program for simulating and reconstructing a slice. We are
starting to write the IDL program for visualization.
We will begin the fan beam project only after the parallel beam
project is completed.
Why a Super Computer:
Aside from the multi processor aspect of this project we need a
super computer to handle the massive calculations and the large volume of
data. One regular size set of data is over a gigabyte. A standard object
requires four thousand calculations involving up to three hundred and
sixty pieces. Each calculation requires each piece to be integrated in
and filtered. There is no practical way to run this program on a small
machine. One data set takes about thirty minutes on a single processor.
For a interactive process this is a unacceptably long time.
Team Members
Sponsoring Teachers
Project Advisor(s)