Samuel Payne and Dr. Bryan Morse, Computer Science

My ORCA research topic was 3D shape representation. I proposed research to develop a new method for representing surfaces that combined two current methods. The oldest method (thinplate spline model) was developed by Turk and O’Brien in 1998. The second method (compactly supported RBF) was developed by my mentor, Bryan Morse in May 2001. These methods have complimenting strengths and weaknesses. Therefore I planned to hybridize these two methods to preserve the strengths in both, and thus overcome their respective weakness.

The goal of the project was to reduce the size of 3D image files, and at the same time maintain usability for 3D modeling software. Medical imaging requires hyper accurate representations. When modeling a human organ, it is not uncommon for the file to have over 1 million data points. This is simply too large for today’s computers. They cannot keep all that information in RAM and manipulate and display it.

With the ORCA grant, I was able to research with Bryan Morse and implement a preliminary version of my algorithm. Although results vary between different files, my program succeeded. It dramatically reduces the size of image file. On average my program reduced the file size by 90%. There were many files reduced by more than 99%, but a few where only 70% reduction was achieved. By reducing the size of the file by an order of magnitude, my program enables computers to deal with data sets that otherwise could not be handled for another 5 years. (Approximately every 5 years, the size of RAM in a computer grows by an order of magnitude.)

This new program has definite advantages over both of its contributing predecessors. It is more advantageous than the thin-plate method, because the size decrease as noted above. It is more advantageous than the compactly supported system because the surface function is defined everywhere, which allows 3D plotting and display programs to use it.

My project was not fully finished but it is a working model, and a basis for further research. While it works on the data sets that are available for testing, currently it requires too much fine tuning for each data set to be universally usable. This research and continuing improvement will be picked up by Bryan Morse.

I would like to thank ORCA for giving me the grant, which enabled me to participate in original, cutting edge research and development. I believe that this research is a contributing factor for my admittance to my current PhD program at the University of California at San Diego. It has given me confidence in my abilities to learn and accomplish. Thank you.