Thank Zoo for the data

This post will be about two things: firstly, giving a big thank you to all those who have helped our project over the past few months; and secondly, providing a summary of where we are going next.

For those unfamiliar with Map the Planets Project, we’ve been developing software to automate the process of analysing planetary images. We’ve been working on statistical methods for making quantitative measurements, such as estimating surface areas, and also predictions for how accurate such measurements can be computed. We’ve been getting very close to achieving our goals on simulated Martian images and we’re hoping to soon have more encouraging results from crater counting on the Moon. We created simulated Martian images using data from HiRISE, but we’ve needed help gathering images and ground-truths for experiments on lunar craters. This brings us to the first item on this post’s agenda, that of thanking all those who assisted in providing test crater data…

Students hard at work using ImageJ software to annotate sample images.

Students hard at work using ImageJ software to annotate sample images.

Thank you Moon Zoo users! We’d like to give a big thank you to all those citizen scientists who have dedicated many hours over recent months counting all the craters around the Apollo 17 landing site. These are the craters we’re using to test our software.

Thank you Moon Zoo team members, especially Roberto Bugiolacchi, Ian Crawford and Katie Joy who have provided us with raw data, lunar images and advice on what exactly planetary scientists would like to use automated systems for.

Undergraduates at the School of Earth, Atmospheric and Environmental Sciences dedicated their afternoons to helping identify Apollo 17 site craters.

Undergraduates at the School of Earth, Atmospheric and Environmental Sciences dedicated their afternoons to helping identify Apollo 17 site craters.

Thank you Manchester University undergraduates, Dayl Martin, Sean Corrigan, Alex Griffiths, Hazel Blake, Gosia Sliz, Joe Scaife, Pavel Kamenov and especially Tim Gregory, who spent their afternoons marking up every crater in sample regions which we’ll be using to calibrate our software. And thanks to Beth Marshall for baking the cookies.

Academic resources are stretched and everybody has a very busy schedule, so we’re hugely grateful for all the assistance we’ve received.

Moving on to the second item on this post’s agenda, we’re planning to do the following things between now and the end of the year…

Mars from Viking Orbiter. Image from NASA.

Mars from Viking Orbiter. Image from NASA.

It’s PhD thesis writing time and this has highlighted the need to repeat a few Martian image experiments. There are still some unexplained effects that are making our Martian surface area measurements less accurate than we would have predicted. We need to test for these discrepancies and correct for them if possible. If we can fix these problems then it will become possible to automatically measure the surface area of all sorts of terrains, such as measuring how much of Mars is covered in dunes for example. Once the statistical theory and image encoding scheme is fully understood there could be hundreds of future applications for our methods. But for now, we just need to understand them well enough to write some key thesis chapters as a proof of concept. The plan is to have this work done by the end of June.

A template crater extracted by our software via images taken from NASA's Lunar Reconnaissance Orbiter.

A template crater extracted by our software via images taken from NASA’s Lunar Reconnaissance Orbiter.

From July we’re hoping to demonstrate our method’s practical utility by automatically generating quantitative size-frequency distributions (SFDs) for impact craters around the Apollo 17 landing site. An SFD plots the number of impact craters in an area against crater sizes. These are useful for estimating the absolute age of a surface, or placing neighbouring surfaces into relative chronological order. To do this we’ll be using a combination of Moon Zoo data and expert annotations. We will measure our success by comparing the accuracy of our crater counts to the accuracies predicted by our error theories, and also through comparison to expert SFDs and Moon Zoo users. This requires a good understanding of the types of noise present in data, as explained in a previous post.

We also have a couple of potential journal papers we’re trying to get published, and perhaps we’ll get to go to a conference or two. Latest news and developments can be found on our facebook page and technical notes can be found at www.tina-vision.net for those interested in the maths.

Advertisements

About pauldtar

Post-doc researcher at the University of Manchester. Working on software to automate the process of analysing images of planets.
Gallery | This entry was posted in Background Science, Space and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s