2 Million Galaxies; % of sky? (APOD 07 Oct 2007)

starnut · Post by **starnut** » Mon Oct 08, 2007 1:08 am

http://antwrp.gsfc.nasa.gov/apod/ap071007.html

How did they count the 2 million galaxies in the image? Did they count in a small square and extrapolate to the rest of the image?

Chris Peterson · Post by **Chris Peterson** » Mon Oct 08, 2007 3:54 am

starnut wrote:How did they count the 2 million galaxies in the image? Did they count in a small square and extrapolate to the rest of the image?

No, they actually counted them individually. Astronomical images are routinely analyzed with special software tools that extract a list of all objects, and which can distinguish stars from extended sources.

The original data for this galaxy survey came from photographic plates, which were digitized with an impressive laser scanner and subsequently analyzed.

geckzilla · Post by **geckzilla** » Mon Oct 08, 2007 12:32 pm

This is a little off topic, but I'm curious to see if you know anything about this, Chris.

http://galaxyzoo.org/

There is a huge image they seem to use with lots of galaxies on it but it seems as though the image has multiple levels of jpeg artifacts on it. Instead of having the huge source image as an uncompressed and therefore much clearer image, they have it stored as a jpeg and then again when the website script pulls the little chunk of the image they want you to look at, interpolates and even rotates it, and then saves it once again into a jpeg, creating another level of artifacts.

But later on it occurred to me that maybe the artifacts I thought I was looking at may have been caused in the original digitizing of the image. I decided to look at a random RAW image I took with my camera which supposedly is uncompressed and therefore should not have any artifacts... but I was very surprised to see that close up, it does have some blurry boxes very similar to jpeg compression artifacts.

Anyway, it's only important in this particular instance because sometimes very small (er, distant, whichever) galaxies are zoomed to very high levels and at that point the compression artifacts can easily be mistaken as arms or randomness that's not actually there.

Chris Peterson · Post by **Chris Peterson** » Mon Oct 08, 2007 4:57 pm

geckzilla wrote:There is a huge image they seem to use with lots of galaxies on it but it seems as though the image has multiple levels of jpeg artifacts on it.

I looked at a number of images served by Galaxy Zoo, and didn't see anything I'd call an obvious compression artifact. I don't think images are rotated; the only processing they seem to apply when the image is served is scaling, which undoubtedly can create some artifacts. The image you linked is odd, but doesn't appear to me to have JPEG artifacts, since the boxy areas are of different sizes.

geckzilla · Post by **geckzilla** » Mon Oct 08, 2007 5:51 pm

That's the thing, I'm not exactly sure, since I haven't ever seen for sure non-lossy images to compare. What they look like are interpolated jpegs to me, but I can't be certain. I decided to make a simulated image, just to try to illustrate my point, which is that any artifacts can distort small specks and confuse unwitting viewers into thinking they are seeing something they are not.

I had thought of it as an entirely preventable thing by not allowing generations of jpegs interpolation to take over the image. I even sent them an email pointing out this apparent flaw but am having regrets about it for not figuring that maybe the artifacts came to exist when the image was digitized or even during the photographic process somehow. Since I have decidedly absolutely 0 insight into what methods they might be using, I thought maybe you would.

Here is what I thought was happening, which I guess I am having a little bit of a hard time explaining... this is a PNG, with jpeg images saved with in, so it should be a true representation albeit a simulation of what I am thinking about. The little ball is a dithered radial gradient with some random noise applied which is something that happens with all low-light photography, and so assume happens with astrophotography as well.

The top frame represents my guess at how an original may have appeared at 500% with no interpolation.

The second frame is the first frame saved at its original size as a jpeg and then interpolated up after being converted to a jpeg.

The third frame was then rotated and saved yet again as a jpeg.

It looks a little different from the images I see on galaxy zoo. I could probably push it further with some variations but I don't want to try and force it when I really don't have enough information to go on.

Galactic Groove · Post by **Galactic Groove** » Tue Oct 09, 2007 7:44 pm

your screen or monitor can affect the quality of the image being displayed. that may or may not help in explaining things though lol

Zednine · Post by **Zednine** » Wed Oct 10, 2007 4:31 pm

I am wondering, what percent of sky does this image represent? I tried to figure it out, but the result doesn’t seem to make sense to me. This is what I did:

The height of the image is about 71% of the width of the image. If the width of the image is about 100 degrees, then the height of the image is about 71 degrees.

If the measure of the whole sky can be expressed by the equation 360 degrees times 360 degrees, then the whole sky can be represented by the number 129,600.

The size of the image can then also be expressed by the equation 100 degrees times 71 degrees, and can be represented by the number 7,100.

If all this is correct, then the percentage of sky that this image represents can be expressed by the equation 7,100 divided by 129,600. This would make the percentage of sky represented about 5.5%.

Is all of this correct? It seems like a 100 degree by 71 degree area would be a lot larger than 5.5% of the sky. Did I figure this out correctly?

Thx, Zednine

geckzilla · Post by **geckzilla** » Wed Oct 10, 2007 5:37 pm

I have seen another description of the image say it is actually about 30 degrees across:
http://map.gsfc.nasa.gov/ContentMedia/990047b.jpg

Dunno which is correct.

Chris Peterson · Post by **Chris Peterson** » Wed Oct 10, 2007 10:45 pm

Zednine wrote:I am wondering, what percent of sky does this image represent? I tried to figure it out, but the result doesn’t seem to make sense to me. This is what I did:

The height of the image is about 71% of the width of the image. If the width of the image is about 100 degrees, then the height of the image is about 71 degrees.

So far, so good. This matches nicely with the APM website description of the image stating it covers about 7000 square degrees.

If the measure of the whole sky can be expressed by the equation 360 degrees times 360 degrees, then the whole sky can be represented by the number 129,600.

There are about 41,253 square degrees covering the sky (look up the formula for the surface area of a sphere).

7000 / 41,253 = 0.17, meaning this image covers 17% of the sky. The APM website says "roughly a quarter of the whole sky" - very roughly, I'd say.

Vision · Post by **Vision** » Thu Oct 11, 2007 11:43 am

I would look at it as, how many galaxies in that small amount of space can we not see? =)

Wadsworth · Post by **Wadsworth** » Thu Oct 11, 2007 5:09 pm

Does anybody know where one can aquire a high resolution version of the APM galaxy map?

Chris Peterson · Post by **Chris Peterson** » Thu Oct 11, 2007 5:30 pm

Wadsworth wrote:Does anybody know where one can aquire a high resolution version of the APM galaxy map?

I assume the images linked at the APM site are the highest resolution images available. The image was made by binning the scanned area into 0.1 degree squares and counting the galaxies in each, which works out about right for the image size given 0.1 degrees per pixel.

Starship Asterisk*

2 Million Galaxies; % of sky? (APOD 07 Oct 2007)

2 Million Galaxies; % of sky? (APOD 07 Oct 2007)

Re: APOD 10/7/07 - 2 million galaxies

2 million galaxies, What % of sky

Re: 2 million galaxies, What % of sky