Citizen Scientists Help Explore the Moon 60
Posted
by
kdawson
from the given-enough-eyeballs-all-craters-are-shallow dept.
from the given-enough-eyeballs-all-craters-are-shallow dept.
Pickens writes "NPR reports that NASA's Lunar Reconnaissance Orbiter is doing such a good job photographing every bit of the moon's surface that scientists can't keep up, so Oxford astrophysicist Chris Lintott is asking amateur astronomers to help review, measure, and classify tens of thousands of moon photos streaming to Earth using the website Moon Zoo, where anyone can log on, get trained, and become a space explorer. 'We ask people to count the craters that they can see ... and that tells us all sorts of things about the history and the age of that bit of surface,' says Lintott. Volunteers are also asked to identify boulders, measure the craters, and generally classify what is found in the images. If one person does the classification — even if they're an expert — then anything odd or interesting can be blamed on them. But with multiple independent classifications, the team can statistically calculate the confidence in the classification. That's a large part of the power of Moon Zoo. Lintott adds the British and American scientists heading up the LRO project have been randomly checking the amateur research being sent in and find it as good as you would get from an expert. 'There are a whole host of scientists ... who are waiting for these results, who've already committed to using them in their own research.'"
A better use of the public's time... (Score:5, Interesting)
...would be to use the statistically-validated user input in a feed-forward image recognition neural network utilizing error feedback that would "learn" to identify the various features of interest. Use edge detection to identify the features of interest (for instance, by number just like a paint-by-number canvas), and have users "identify" what they see. We're talking about invariant scale here, which vastly simplifies the learning process as well as automated feature measurement.
I was doing this in the '90s using multi-band spectral imagery from LANDSAT with good success. I would imagine there have been some advances in this area since that time.
Re:Mutual Benefit (Score:3, Interesting)
Because it is basic game theory? You want the little hamster to continue running around the little wheel you give him a cookie to work for. If he gets little nibbles of the cookie he'll work HARDER trying to get more cookie, thus giving you more work. Hell nobody is saying they have to give them the actual recorded data in real time, just throw the monkey a reward for pushing the button. Maybe something that ONLY shows how you are doing? Surely that would discourage the cranks while giving the hamster a reason to keep running the wheel.
For those that would like to do a little amateur astronomy themselves, or just haven't heard about it and would like to try some really cool FOSS, give Stellarium [sourceforge.net] a go. The amount of detail it gives is just insane, with my retired NASA engineer neighbor actually able to plan his viewings of Jupiter's moons by running Stellarium simulations beforehand to allow him to find the best times for viewing in our area. It works on Linux, OSX and Windows, is free, and will actually run decently on 5 year old laptops. I'll have to see if the Stellarium guys would like a copy of his latest project, which is using Stellarium to give a visual representation of the objects he is doing spectral analysis of. Stellarium really gives a great presentation and helps the non astronomers to understand what the college astronomy club is doing and the college kids really love it.
If any of the Stellarium guys read this, great work guys. Your software really is top notch and professional. Oh and he is giving full credit to Stellarium for the visuals, as well as making sure the website is printed on all material. He also made a really nice set of graphic art CD covers for Stellarium, which I'll have to find a place to upload.