Press Kit

What is Astrometry.net?

Astrometry.net is a group of scientists whose goal is to

organize, annotate and make searchable all the world's astronomical information.

Our main project is a blind astrometry solver, a program that looks at images of the night sky and determines where the telescope was pointing and which stars are contained in the image. Stargazers do the same thing when they look up at the night sky and find the constellations, but our program can also recognize tiny images that cover one ten-millionth of the sky, containing no stars visible to the unaided human eye.

Our software is called a blind astrometry solver: blind means that it requires no extra information about the image; it just looks at the image itself. The word astrometry (astro = the stars + metry = measurement) refers to the branch of astronomy concerned with measuring the positions, motions, and brightnesses of celestial objects. When astronomers are analyzing a new image, one of their first tasks is often to solve the astrometry, which means placing the image in a standard reference frame so that it can be compared with, and described in terms of, other images and observations. Essentially, solving the astrometry of an image means determining how pixel positions in the image map to positions on the sky.

Our blind astrometry solver automates this process: you give us a picture, we tell you how to map between pixels and positions on the sky.

Once we have solved the astrometry of an image, it is easy to label your image, showing where the constellations, nebulae and other "interesting" objects are. It is also possible to show you images that other people have taken of the same region of the sky.

What is the status of the project?

We have launched a web version of our system, and we have invited several dozen professional and amateur astronomers to be "alpha-testers". We have solved thousands of images uploaded by our testers. We have also released our code so that other individuals and groups can adapt it to their purposes.

We are currently working on ways of using the images we have solved. For example, we are building a browseable map of the sky by combining all the images we have solved into a high-resolution map that is constantly being updated as users contribute more images.

How does it work?

We start with a large catalog of star positions. We are currently using the USNO-B1.0 catalog, which was created by the US Navy Observatory and contains the positions and brightnesses of about one billion stars. We take this catalog of stars and find a large number of "skymarks" (landmarks for the sky). Each skymark is composed of four stars, and the skymark describes the relative positions of these four stars. We need to build enough skymarks that each image we try to solve will contain at least one, and preferably many, skymarks. We call this big collection of skymarks an index - an index in a book maps from a word to the places in the book where that word can be found; our index maps skymarks to the places on the sky where that skymark can be found.

When you submit an image, we first run some image processing steps to find stars. Next, we start looking at sets of four stars in your image. For each set of four stars, we see if it matches one of the skymarks in the index. Often, one skymark can be mistaken for another, so after we find a skymark that seems to match, we ask, "if this skymark really is a match, where else would we expect to see stars in this image?" If a lot of the predicted stars really are there, then the match must be correct and your image is solved.

Let's see some examples!

This image is copyright Russell Croman,  http://www.rc-astro.com/ . It was  Astronomy Picture of the Day, 2003 Feb 28

Here are the stars we found in the image:

Here is the skymark that we matched. The green circles are stars in our index - notice how many red and green circles overlap - this is definitely a correct match.

Once the image has been correctly matched, we can easily label interesting objects in the image:

Who is working on Astrometry.net?

The team includes:

• Sam Roweis (Associate Professor, Computer Science, University of Toronto) - co-Principal Investigator
• David W Hogg (Associate Professor, Physics, New York University) - co-Principal Investigator
• Dustin Lang (PhD candidate, Computer Science, University of Toronto)
• Keir Mierle (Electrical Engineering, University of Toronto)
• Jon Barron
• Christopher Stumm