The Take-Away with Jacob Moul

May 1, 2016

Over the past couple of weeks I have concluded all of my testing and am left with a fair amount of data. 

What Does It All Mean??

Running all the algorithm combinations resulted in lots of control networks (files that contain all the match points between the two images). After looking through all these networks, I have determined that one feature detector algorithm finds a lot of points (in the 5000s), and the other two detectors both find significantly less (in the 200s). In addition, each detector-extractor algorithm resulted in distinct clusters in different areas on the Moon's surface. I will have to do more analysis to see if the effectiveness of each combination was due to the level of each match condition (geometry, lighting, overlap), but at the moment, there is no clear pattern.

What these results tell us is that more work needs to be done, both in developing the code and in testing the code to determine the problems. My advisor told me something the other day that I did not realize: all of the algorithms we've been using have been developed for use on images of the Earth's surface, and have just been used on extraterrestrial images out of convenience. Due to the sometimes extreme difference between terrestrial and extraterrestrial surfaces, the algorithms will occasionally fail.

Fortunately, my project has been able to identify some of the issues that these algorithms run into (i.e. clustering and small numbers of match points). The purpose of my project was to help the USGS improve their image matching software. Because there was no concrete goal involved in that purpose, I would consider my project to be moderately successful. Even though I don't have the technical knowledge to address the problems myself, just identifying them provides a sort of roadmap for developers at the USGS to follow.

What Did I Learn?

Computers are frustrating. Debugging took up a good third of my project time, and that was a pain in the butt. Especially because I only have a basic understanding of programming so it was extremely difficult to identify and fix the problems. Luckily, my advisor did most of the heavy lifting in this respect...

I also learned that space is super cool. I already knew this to a certain extent, but staring at pictures of the Moon all day really sparked my interest in space. During my project I started reading a book about cosmology (the study of the universe) and dark matter and things like that and I think it's extremely interesting. Right now I'm declared as mechanical engineering major, but who knows, maybe I'll end up studying astrophysics instead. OR maybe I'll double major. We'll see what happens.

In all seriousness though, the very basic programming skills I learned during the course of my project will provide a good base to build off of in the future. I got a lot out of my project and I think it definitely has shaped the academic path I want to pursue in college.

To all who read my blog, thank you for keeping up with my project and for being interested!

Jacob Moul