Finding Asteroids

by Joe Heafner
Catawba Valley Community College, Hickory, NC

Finding asteroids in SDSS images can be tedious when done by simple visual inspection of the images. If you've done the Asteroids project, you've looked through lots of images to find asteroids like the one shown to the right, and you've seen how hard you have to look. A much more efficient way it to use computer software to help find asteroids.

Traditionally, this involves taking an image of the same part of the sky at no fewer than two different instants and then using computer software to "stack" the images, which simply means to overlay the two images so that corresponding stars are aligned. The computer software then rapidly displays the individual images in succession, a technique called "blinking". The stars, being fixed objects, do not wiggle but asteroids, being moving targets, will appear to wiggle from image to image.

SDSS images allow us to use a slight variation on this technique that relies on the fact that an image of a given part of the sky is taken with different filters at slightly different times. Unlike optical images, in which asteroids show up with the same color from image to image, SDSS images show asteroids in two different colors. An SDSS image is also a composite of three different colored images, and each separate component image is taken at a slightly different time from the other images. Think of an SDSS image as three different images that have been digitally "stacked" as described above. If we could somehow "unstack" an SDSS image and then "blink" the individual component images, we might just be able to detect asteroids! In practice, this technique works and here is how to do it.

The first step is to obtain a computer program that will manipulate the images for us. An excellent choice is ImageJ, developed by the National Institutes of Health. ImageJ is a Java program that runs under Linux, Mac OS 9, and Mac OS X, and Windows. It is freely available for download at the ImageJ website. Your computer must have the ability to run Java programs. Your instructor or teacher can give you more information about making sure your computer can run Java programs - some additional software may be necessary if not already present.

ImageJ has many advanced features needed for different kinds of image analysis, including astronomical work and medical work. If you have not done so, or if your instructor has not yet done so, you will need to download ImageJ and install it on your computer. Instructions are provided in the download archive.

The first step, assuming ImageJ is now properly installed on your computer, is to get an SDSS image. You can use the Get Fields tool to retrieve an image given its run, camcol, and field numbers. If you don't have a special run, camcol, and field that you want to work with, you can just choose these numbers at random. Be sure to get the largest image you can, 1984x1361 pixels. This will make seeing the asteroids visually easier.

Once you have an image, right-click on it to save the image to your desktop. We recommend that you rename the image according to its run, camcol, and field. The example image we are using has been named r752c1f373.jpg, indicating Run 752, Camcol 1, Field 373. Note that the image has been saved in JPG format, which is fine for our purposes.

Sample image: Run 752, Camcol 1, Field 373 (click to open in a new window)

The second step is to load the image into ImageJ for analysis. Your instructor or teacher will give you instructions on how to start the ImageJ program. Once it has been started, select File -> Open from the program's menu and load your image into ImageJ. The image will then be displayed in a window on your screen. The screenshot to the right shows what a sample image looks like opened in ImageJ.

The third step is to separate the composite image into its three component images. The composite image is called an RGB image, meaning a combination of red, green, and blue images. Actually, these are just names given to the three component images and they might not necessarily correspond to actual red, green, or blue as we usually think of them. This is not important for our purposes however.

To do this in ImageJ, select Image -> Color -> RGB Split. You will see the original image replaced by three separate images. Note that each one will be labeled with its component color. A screenshot of the split image is shown to the left.

The fourth step is to create an image "stack" with the three component images. To do this, select Image -> Stacks -> Convert Images to Stack. The screenshot of the stacked image is shown to the right.

We now have one single image again, or so it appears. In a way, we are just undoing the separation we just performed but that is not exactly what we are doing. We are instructing ImageJ to put the component images together in a way that allows us to animate them, blinking them in quick succession.

The fifth step is to animate the stack to see what wiggles. To do that, select Image -> Stacks -> Start Animation. By default, ImageJ will cycle through the three component images at one second intervals, but this interval can be adjusted if desired. Depending on your computer's speed and memory capacity, blinking may slow down your computer and it may be slow to respond to key and mouse movements. A better option might be to use the > and < keys to manually cycle through the component images so you can start and stop as you wish. Stars and galaxies in the images will remain stationary with respect to each other. Their brightnesses may change, but their positions will not change. If you see a dot that moves as ImageJ blinks the images, you have most likely found an asteroid! The screenshots below show two blinking images. The image below show the location of an asteroid (marked in yellow).

You can get the location of your find in the image by moving the mouse cursor to the dot that moves. The coordinates of this dot, in pixels, will be displayed in the main ImageJ window, the one with the tool buttons. In our sample image, a moving object is seen at coordinates X=894, Y=225. You should always look at the same coordinates in the original image, which you can reload into ImageJ using File -> Open again, to confirm whether or not there is actually something there. An asteroid on the original image will show up as two close but separate dots, one usually green and the other usually blue.

In accordance with good scientific practice, your asteroid discoveries should always be followed up by visual inspection of the original image. When you record your findings, be sure to include the coordinates of each asteroid in your records so that other astronomers can verify your findings and use them for future work and for reference. Upon looking at the original image, we surely do see a pair of blue and green dots at very close to the location above. You can always zoom the image to get more precise coordinate readings.