6 ms·

The whole aspect of facial recognition upon satellites will sure open up some messy legal cases in the future I feel.

by Zenst 6y ago

walrus01 6y ago
Unless something dramatic changes in the field of optics, lenses and how to deal with atmospheric blur at those distances, resolving features smaller than 15-20cm from low earth orbit remains a low probability. Take a look at some youtube videos of a Nikon P1000 at maximum zoom looking through ordinary atmosphere at some distance away for an example. https://www.youtube.com/watch?v=LhQlwKX3LQA https://www.youtube.com/watch?v=LhQlwKX3LQA
- IceWreck 6y ago
  > Take a look at some youtube videos of a Nikon P1000 at maximum zoom looking through ordinary atmosphere at some distance away for an example. https://www.youtube.com/watch?v=LhQlwKX3LQA https://www.youtube.com/watch?v=LhQlwKX3LQA This is unbelievable.
  - giantrobot 6y ago
    While that's an impressive zoom, the camera is moving much slower relative to the targets (~0km/s) than if it were in even a low orbit. It's also imaging people at the extent of its zoom at only a few km away. In orbit it would be hundreds of km from its targets. CubeSats and other small satellites can capture useful imagery but they're not spy satellites. There's optical limits to their capabilities because of their size.
- ceejayoz 6y ago
  I think the more likely use is supplementing existing facial recognition. Recognize a face on a security camera, then be able to track their movements out and about after that. Before, too, if you store the satellite footage. https://www.buzzfeednews.com/article/christianstork/spy-planes-over-american-cities https://www.buzzfeednews.com/article/christianstork/spy-plan..., but at scale.
  - walrus01 6y ago
    Pattern of life analysis correlated with ground based intelligence, where your car goes, where it usually parks in the day and at night, etc. You would need a lot of LEO satellites to approach anything like persistent coverage, however, current sun synchronous things have very long revisit times for any particular site.
    - JulianMorrison 6y ago
      Also clouds kind of get in the way of anything optical - unless it's a drone lower than the cloud level.
- londons_explore 6y ago
  Atmospheric distortions are worse when looking horizontally though the atmosphere (think marriage on the horizon). It's also only the bottom few miles of atmosphere that is really thick. Looking vertically from 400km up, there actually isn't too much distortion. The limiting factor is making high enough quality lenses and getting enough light onto your sensor without too much diffraction.
  - pintxo 6y ago
    But hard to get a face shot from a 90 deg angle, as people tend to focus their attention on the ground rather the sky.
    - saberdancer 6y ago
      You don't need to look from 90 deg angle. Satellites often times scan ahead/behind so they don't always scan at 90 deg angle. Look at this image that is probably from a spy satellite. It's almost at 45 degree angle. https://www.thedrive.com/the-war-zone/29634/trump-tweets-intelligence-image-after-irans-rocket-explosion-and-offers-them-best-wishes https://www.thedrive.com/the-war-zone/29634/trump-tweets-int...
      - pintxo 6y ago
        The previous post argued that the influence of the air is less severe looking straight down. If we now go back to looking at an angle, we are back to square one, the problem of distortion due to the amount of air between the observer and the observed.
    - ta988 6y ago
      If you need that you can send a drone though.
  - andreareina 6y ago
    If I'm running the numbers right, then using the Hubble's maximum theoretic angular resolution of 2.8e-7 rad[1], something 400km up would still not resolve features smaller than 11.2cm. That's the absolute best, diffraction-limited case. Don't know how close we are to that on real-world optics. [1] https://opentextbc.ca/universityphysicsv3openstax/chapter/circular-apertures-and-resolution/ https://opentextbc.ca/universityphysicsv3openstax/chapter/ci...
- helsinkiandrew 6y ago
  Presumably, orientation is also an issue - satellites will tend see tops of heads more than faces full on.
  - jl6 6y ago
    The obvious solution: pivot face-recognition algorithms into top-of-head-recognition algorithms.
    - yuliyp 6y ago
      People can comb their hair quite differently
james-bcn 6y ago
Do you need facial recognition when nearly everyone carries a mobile phone?
- SamuelAdams 6y ago
  Not to mention RFID is being added to state ID's, driver's licenses, and passports. [1]: https://www.dhs.gov/enhanced-drivers-licenses-what-are-they https://www.dhs.gov/enhanced-drivers-licenses-what-are-they
  - gruez 6y ago
    You mean the ones that can only be read from a few feet away at most, and can be blocked by putting them in easily accessible "rfid blocker" wallets?
    - deepgrave4 6y ago
      How many people are out buying RFID wallets to protect themselves from spy devices they don't know exist?
JulianMorrison 6y ago
Faces no, as other people have said - except, what about integrated systems that combine ground based CCTV with satellites, or drones? You might only be able to see someone's head and shoulders, but that's good enough to follow them as they walk through a crowd, and you can use that to decide which CCTV to use. Presuming it's all networked and accessible, but that might be the case at some future point.
- EdwardDiego 6y ago
  I wonder how gait recognition from space would work, if it all.
  - JulianMorrison 6y ago
    Feet would be single pixels if they were visible at all. And they don't project much in front or behind when walking.
simonh 6y ago
The highest resolution images we know are possible in theory max out at around the 10cm level due to atmospheric effects. It seems we can get quite close to this with the big spy satellites the Hubble Space Telescope was based on. Fortunately we have a really great example available to examine. https://www.npr.org/2019/08/30/755994591/president-trump-tweets-sensitive-surveillance-image-of-iran https://www.npr.org/2019/08/30/755994591/president-trump-twe...
- baybal2 6y ago
  It's known that the latest spy satellites have some kind of folding mirrors, and adaptive optics - things not being sci-fi anymore for a few decades at least.
- ramraj07 6y ago
  The “theory max” is not that hardcoded, superresolution is a field and we don’t know what all they are up to with billions to burn. There were reports at some point about high altitude drones stationary above cities that can read license plate level data (not license plates themselves) at video rates over the entire city so who knows!
duxup 6y ago
Or just another state that you have no legal recourse in doing whatever they want. How is your foreign social credit score anyway?

The whole aspect of facial recognition upon satellites will sure open up some messy legal cases in the future I feel.

by Zenst 6y ago

The whole aspect of facial recognition upon satellites will sure open up some messy legal cases in the future I feel.

Unless something dramatic changes in the field of optics, lenses and how to deal with atmospheric blur at those distances, resolving features smaller than 15-20cm from low earth orbit remains a low probability. Take a look at some youtube videos of a Nikon P1000 at maximum zoom looking through ordinary atmosphere at some distance away for an example. https://www.youtube.com/watch?v=LhQlwKX3LQA https://www.youtube.com/watch?v=LhQlwKX3LQA

IceWreck 6y ago
> Take a look at some youtube videos of a Nikon P1000 at maximum zoom looking through ordinary atmosphere at some distance away for an example. https://www.youtube.com/watch?v=LhQlwKX3LQA https://www.youtube.com/watch?v=LhQlwKX3LQA This is unbelievable.
- giantrobot 6y ago
  While that's an impressive zoom, the camera is moving much slower relative to the targets (~0km/s) than if it were in even a low orbit. It's also imaging people at the extent of its zoom at only a few km away. In orbit it would be hundreds of km from its targets. CubeSats and other small satellites can capture useful imagery but they're not spy satellites. There's optical limits to their capabilities because of their size.
ceejayoz 6y ago
I think the more likely use is supplementing existing facial recognition. Recognize a face on a security camera, then be able to track their movements out and about after that. Before, too, if you store the satellite footage. https://www.buzzfeednews.com/article/christianstork/spy-planes-over-american-cities https://www.buzzfeednews.com/article/christianstork/spy-plan..., but at scale.
- walrus01 6y ago
  Pattern of life analysis correlated with ground based intelligence, where your car goes, where it usually parks in the day and at night, etc. You would need a lot of LEO satellites to approach anything like persistent coverage, however, current sun synchronous things have very long revisit times for any particular site.
  - JulianMorrison 6y ago
    Also clouds kind of get in the way of anything optical - unless it's a drone lower than the cloud level.
londons_explore 6y ago
Atmospheric distortions are worse when looking horizontally though the atmosphere (think marriage on the horizon). It's also only the bottom few miles of atmosphere that is really thick. Looking vertically from 400km up, there actually isn't too much distortion. The limiting factor is making high enough quality lenses and getting enough light onto your sensor without too much diffraction.
- pintxo 6y ago
  But hard to get a face shot from a 90 deg angle, as people tend to focus their attention on the ground rather the sky.
  - saberdancer 6y ago
    You don't need to look from 90 deg angle. Satellites often times scan ahead/behind so they don't always scan at 90 deg angle. Look at this image that is probably from a spy satellite. It's almost at 45 degree angle. https://www.thedrive.com/the-war-zone/29634/trump-tweets-intelligence-image-after-irans-rocket-explosion-and-offers-them-best-wishes https://www.thedrive.com/the-war-zone/29634/trump-tweets-int...
    - pintxo 6y ago
      The previous post argued that the influence of the air is less severe looking straight down. If we now go back to looking at an angle, we are back to square one, the problem of distortion due to the amount of air between the observer and the observed.
  - ta988 6y ago
    If you need that you can send a drone though.
- andreareina 6y ago
  If I'm running the numbers right, then using the Hubble's maximum theoretic angular resolution of 2.8e-7 rad[1], something 400km up would still not resolve features smaller than 11.2cm. That's the absolute best, diffraction-limited case. Don't know how close we are to that on real-world optics. [1] https://opentextbc.ca/universityphysicsv3openstax/chapter/circular-apertures-and-resolution/ https://opentextbc.ca/universityphysicsv3openstax/chapter/ci...
helsinkiandrew 6y ago
Presumably, orientation is also an issue - satellites will tend see tops of heads more than faces full on.
- jl6 6y ago
  The obvious solution: pivot face-recognition algorithms into top-of-head-recognition algorithms.
  - yuliyp 6y ago
    People can comb their hair quite differently

james-bcn 6y ago

Do you need facial recognition when nearly everyone carries a mobile phone?

SamuelAdams 6y ago
Not to mention RFID is being added to state ID's, driver's licenses, and passports. [1]: https://www.dhs.gov/enhanced-drivers-licenses-what-are-they https://www.dhs.gov/enhanced-drivers-licenses-what-are-they
- gruez 6y ago
  You mean the ones that can only be read from a few feet away at most, and can be blocked by putting them in easily accessible "rfid blocker" wallets?
  - deepgrave4 6y ago
    How many people are out buying RFID wallets to protect themselves from spy devices they don't know exist?

JulianMorrison 6y ago

Faces no, as other people have said - except, what about integrated systems that combine ground based CCTV with satellites, or drones? You might only be able to see someone's head and shoulders, but that's good enough to follow them as they walk through a crowd, and you can use that to decide which CCTV to use. Presuming it's all networked and accessible, but that might be the case at some future point.

EdwardDiego 6y ago
I wonder how gait recognition from space would work, if it all.
- JulianMorrison 6y ago
  Feet would be single pixels if they were visible at all. And they don't project much in front or behind when walking.

simonh 6y ago

The highest resolution images we know are possible in theory max out at around the 10cm level due to atmospheric effects. It seems we can get quite close to this with the big spy satellites the Hubble Space Telescope was based on. Fortunately we have a really great example available to examine. https://www.npr.org/2019/08/30/755994591/president-trump-tweets-sensitive-surveillance-image-of-iran https://www.npr.org/2019/08/30/755994591/president-trump-twe...

baybal2 6y ago
It's known that the latest spy satellites have some kind of folding mirrors, and adaptive optics - things not being sci-fi anymore for a few decades at least.
ramraj07 6y ago
The “theory max” is not that hardcoded, superresolution is a field and we don’t know what all they are up to with billions to burn. There were reports at some point about high altitude drones stationary above cities that can read license plate level data (not license plates themselves) at video rates over the entire city so who knows!

duxup 6y ago

Or just another state that you have no legal recourse in doing whatever they want. How is your foreign social credit score anyway?