Slashdot Mirror
"Google Satellite" To Be Launched This Week
Lord Satri writes "Well, almost. Google signed an exclusivity deal with GeoEye regarding GeoEye-1, the most advanced high-resolution, civil, remote-sensing satellite to date. This must be annoying for other high-resolution, remote-sensing data users since Google already has an exclusivity deal in place with DigitalGlobe, the other major civil satellite imagery provider. From the CNet article: 'Under the deal, Google is the exclusive online mapping site that may use the imagery... in its Google Maps and Google Earth product. And as a little icing on the cake, Google's logo is on the side of the rocket set to launch the 4,300-pound satellite in six days from Vandenberg Air Force Base in California. Terms of the deal weren't disclosed. GeoEye-1 will orbit 423 miles above Earth, but it will be able to gather imagery with details the size of 41 centimeters... Google, though, is permitted to use data only with a resolution of 50 cm because of the terms of GeoEye's license with the US government.'"
Is a Google satellite evil or not evil? Discuss.
John
Some precisions on my summary. DigitalGlobe is obviously not the only other remote sensing data provider, but it's GeoEye main competitor in civil high-resolution multispectral remote sensing. GeoEye is itself the merging of two other previous major players on the same playing field, OrbImage and Space Imaging.
As for my claim of an agreement between DigitalGlobe and Google, see this two years old entry. The original archive for the DG message is here (the link on /geo does not work anymore).
One of the obvious questions that comes to mind is to which extent these exclusivity deals have negative impacts on other remote sensing imagery customers, small or big.
Another question is; does Google really needs such a deal to provide the best webmapping and virtual globes-related tools?
Animoog.org
50cm restriction? do they have something to hide??
Shiny new browser that can do everything and fancy new satellite. The only thing missing is my new RFID implant.
Take the cheese to sickbay, the doctor should see it as soon as possible - B'Elanna Torres, "Learning Curve"
Within 1-2 years other countries will have civilian spy satellites that break 50cm, putting American companies at a disadvantage.
The USA will have 3 choices:
Shoot the birds down, literally.
Shoot the birds down, politically - bully the other countries into imposing similar limits.
Lower or eliminate the artificial limit.
Anyone remember when encryption software was considered a munition? Apple and other companies had to go through hoops to export it, putting them at a distinct disadvantage over non-American companies.
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
Though at the time.. it was just a April Fools joke..
Guy comes out of bar holding a girl's hand while walking home. Suddenly, a targeted ad for condoms is projected on the ground in front of them.
The AACS key is NOT 0xF606EEFD628B1CA427BEA93A9CA9773F
It means a car shows up as 4 pixels by 12 pixels. The top of your head is part of a single pixel along with a square foot of sidewalk.
Google already has higher-res data for populated areas of several countries from aircraft reconnaisance. The satellites are for everything else.
Unfortunately, there is a physical limit to how good an image taken from 400 miles away can be.
My understanding is that at 50cm resolution, an object that is 50cm across would appear as a single pixel on the image. So, a manhole cover in the street might show up as a single pixel at that resolution. A car's hood might be four pixels, etc. Objects that are smaller than 50cm should not be detectible, especially if they are close to the same color as the background. However, if you zoom in on almost any American city to maximum resolution on google maps's satellite view, you will clearly see traffic lines. Traffic lines are roughly 10cm wide. Often these resolve to two pixels.
So, either my understanding of satellite photo resolution is wrong, or Google can already go to 10cm, and possibly even 5cm resolution.
When our name is on the back of your car, we're behind you all the way!
Defining optical resolution from space is a bit tricky, as several generations of optical engineers have discovered.
The main criterion is the telescope's point spread function - this is roughly the angular diameter that a pinpoint star appears to be, as seen through the telescope. We want the smallest point spread function, and it should map onto about one to three sensor pixels. (arguments go here about over/undersampling).
The Fourier Transform of the point spread function is the Optical Transfer Function, which is a graph of the spatial frequencies response of the telescope. It's analogous to a hifi's frequency response ... it's an engineering challenge to prevent high frequencies from getting rolled off.
The main limit for high resolution is the diameter of the primary mirror (All mirrors and optical elements, no matter how perfect, have diffraction effects which spread out the light and reduce resolution. The bigger the entrance pupil, the greater the resolution) For the GeoEye, orbiting at 684Km and a resolution of 0.4m, I roughly calculate the primary mirror is somewhere around a half-meter diameter or so, depending on the wavelength of light it's optimized for.
Other things limit resolution - scattering of light in clear air (Rayleigh scattering) screws up the image, especially in the blue. Dust, haze, clouds and urban pollution are a bother, but not as much as you might think. Naturally, there's lots of image processing software ... quite compute intensive.
A typical human, seen from above and not casting a shadow, is about 20 to 60 cm across. So someone walking down the street should appear on a few (1 to 5) pixels. Not enough to recognize someone, especially since you're looking down on 'em.
Generally, images taken from aircraft have better resolution (they're closer, and there's less Rayleigh scattering). Perhaps airlines will attach automated, downward looking hires cameras to their daily flights.
the diffraction effect is not the only issue; but its worked-around.
recently, there were 'multiple exposure' (roughly) algorithms being used to 'look thru' the heat, pollution and general waviness of the sky, in plotting out celestial objects.
and even *with* diffraction, you can overcome it with sharpening. I often shoot my photos 'with too high an f-stop' according to common theory; but my post-processing overcomes the diffraction issues in practice; and I get the nice large depth-of-field that I was after with quite good sharpness, as well.
if you get multiple shots, exposures or angles of a subject, you can 'subtract out' quite a lot of noise and distortion. single shots can't do this but multiple ('high dyn range' or HDR) shooting can.
--
"It is now safe to switch off your computer."
This is the american way of doing business. Competition exists to ensure that customers get the best possible price. That's why we tolerate it. That's why we encourage it. When a company talks about putting up "barriers to entry" and signs exclusive deals with all the suppliers, we don't get the benefits of competition anymore.
How we know is more important than what we know.
Given the orbital height of 423 miles above the surface, this satellite will have a near-polar sun-synchronous orbit similar to other land imaging satellites (the Landsats, IKONOS, etc).
vegetables, oh teh horror! ;)
The interesting thing is that in order to get such an orbit, it has to pass over other countries. Will Google take footage of other countries? If so, will it use that footage? That would probably require some intense international negotiations.
Actually, it will not. I'm not sure if it's codified anywhere in international law or just by historical precedent, but a nation's airspace does not extend into space. A satellite can legally take photos of anything it can see, and there's little a country can do about it except hide things under cover or shoot it down (which likely would be considered an act of war).
Some countries (like the US) can exert control in limited ways by restricting operations if the imaging company does business in the country, but that's it.
Google has quite detailed satellite photos of Pyongyang, North Korea - I'm sure they didn't really agree to that.
Worst...sig...ever!
Of all the evil corps predicted to rise in the future by all the cyberpunk and sci-fi writers. I'll bet you not one of them was named "Google"
I thought we already established that WMD are only 49cm across. You don't need to hide it, just paint it beige and it will look like a nude woman sunbathing... or a litter of spaniels. What is the differnce between the 3?
Slowly waving my hand - "This is not the sig you are looking for."
How weird will that be, if 50 years from now people associate the word Google, like they associate the word Nazi?
Then we can all say "back in my day Google was just a search engine, not a military force that had every bit of information on each human on earth. All hail Page Brin."
Can I bum a sig?
Most satellites for earth observation use sun-synchronous orbits. These orbits let the satellite's cameras take pictures ob objects at the same solar time. This means that it will pass overhead at the same local time every day ... so the images will have the same shadow characteristics.
You accomplish this by making the orbit precess exactly 360 degrees per solar year.
These orbits are typically nearly circular, but needn't be; you can put a spy satellite into a sun-synchronous elliptical orbit, so it'll swoops down and photograph near perigee, then waste a lot of time around apogee.
Since this orbit is around 684 Km, it can be shown that it must be pretty close to circular, has an orbital period of around 100 minutes, and its inclination is probably about 96 to 100 degrees (meaning that the satellite is slightly retrograde - 90 degrees inclination is polar, zero degrees is equatorial) In turn, this means that pretty much all of earth will be seen by the satellite, except for 8 degree circles around the poles.
A 96 minute period means that each successive orbit will look down on a place 15 degrees west ... one time zone to the west.
Geosynchronous orbits are pretty useless for this type of work, since they're so far away (you need really big telescopes to get much resolution). Also, you'd only see one hemisphere, and half the year it'd be nighttime over the areas you want to see.