3 ms·

My understanding is that these satellites are using push broom scanning [1]. So I don't see any meaningful sense in which the different scanners are separated b

by tgb 13y ago

My understanding is that these satellites are using push broom scanning [1]. So I don't see any meaningful sense in which the different scanners are separated by "a few hundred pixels". They're separated by inches, which corresponds to time. For this same reason, I don't see why the altitude of the satellite is important. But I've never really studied this so I'd be happy to be corrected. [1] http://en.wikipedia.org/wiki/Push_broom_scanner http://en.wikipedia.org/wiki/Push_broom_scanner

InclinedPlane 13y ago
They're not mapping the Earth to a resolution of microns though. Does that help make sense of the fact that there isn't a 1-to-1 scale mapping between the Earth and the sensors? Each image sweeps each pixel across a length along the Earth of nearly a meter. If you try to imagine how this maps onto the sensor head it rapidly breaks down. Each pixel seemingly overlaps with other pixels, each line seemingly overlaps with other lines. But that's not what happens, instead there are optics which effect spreading the image plane over a large area on the Earth with a separation of several meters between each color line. Think about the optics in reverse. Imagine turning the satellite's sensor head into a hologram then sending it backwards through the optics until it reaches the Earth, where it would be magnified to an enormous size and each pixel would be the size of a human.
- tgb 13y ago
  I do not understand your description in terms of push broom scanners. Scanning is done by a linear array of sensors that detect at varying angles that go off perpendicular to the angle of motion of the satellite. Imaging in the direction of motion of the satellite is done by physically moving the satellite - which of course is happening quite quickly. This is like looking through a slit in a door frame - you move the light sensor perpendicular to the slit to see the objects inside (except not a perfect analogy: it would be more like carrying the doorframe with you as you move so that you need to move x distance to see something x distance from what you were first seeing). The image is being taken orthographically along one direction. So for the red sensor to see what the green sensor is currently seeing, the satellite must move forward by a few inches so that the red sensor is above the same spot as the green sensor was earlier. This is not how traditional photography works. There is actually a one-to-one correspondence of motion of the sensor with displacement of the imaged object (in one direction). This is similar to how a "photo finish" is done [1] Since the satellite moves much faster than the plane and the displacement between the sensors is small, it seems that the plane should not have traveled very far. [1] http://en.wikipedia.org/wiki/Photo_finish http://en.wikipedia.org/wiki/Photo_finish
  - mannykannot 13y ago
    You are assuming that the lines of sight of each of the arrays are parallel with one another, so that their separation on the ground is the same as the separation of the arrays on the satellite. If they are at an angle to one another, the separation of these lines at the surface will be different from (and possibly in a different order than) the separation of the arrays. It is the surface separation, divided by the speed of the satellite, that determines the time between images of the same point on the surface by different arrays. If the arrays all share the same lens, that would cause this effect.
    - tgb 13y ago
      I'm not sure what you're saying. I think you are either saying the same thing as InclinedPlane (see my response) or are suggesting that either intentionally or due to tolerances in machining or some such, the red and green sensor units are at some small angle from each other. Assuming you're saying the latter, since the distance is large, this small angle contributes a large displacement on the ground. Afterwards the image would be stitched together to compensate for that via some calibration routine which would work for only stationary objects. That is entirely plausible, I hadn't considered that before and had just been assuming that the satellite construction was 'ideal'. This sounds like the most likely reason for such an artifact to occur, though I still wonder whether this particular imperfection would be done given the high level of precision necessary for the entire project.
      - InclinedPlane 13y ago
        It's not due to imprecision, it's due to design and fundamental optical constraints. There's nothing special about the optics used for a push broom imager, it's just an ordinary lens. If you put a full frame sensor at the focal plane you could take an image of a several kilometer square patch of Earth. If instead you put several different lines of sensors in the focal plane and put each behind a color filter then you'd have a multi-spectral push broom sensor. But each line does not view the same part of the Earth at the same time. Because the sensors scan the Earth they end up imaging the same spot within a fairly short delay. But optically it's not possible to have each color line viewing the same horizontal slice of the image of Earth at the same time, that's not the way lenses work.

by tgb 13y ago

InclinedPlane 13y ago

They're not mapping the Earth to a resolution of microns though. Does that help make sense of the fact that there isn't a 1-to-1 scale mapping between the Earth and the sensors? Each image sweeps each pixel across a length along the Earth of nearly a meter. If you try to imagine how this maps onto the sensor head it rapidly breaks down. Each pixel seemingly overlaps with other pixels, each line seemingly overlaps with other lines. But that's not what happens, instead there are optics which effect spreading the image plane over a large area on the Earth with a separation of several meters between each color line. Think about the optics in reverse. Imagine turning the satellite's sensor head into a hologram then sending it backwards through the optics until it reaches the Earth, where it would be magnified to an enormous size and each pixel would be the size of a human.

tgb 13y ago
I do not understand your description in terms of push broom scanners. Scanning is done by a linear array of sensors that detect at varying angles that go off perpendicular to the angle of motion of the satellite. Imaging in the direction of motion of the satellite is done by physically moving the satellite - which of course is happening quite quickly. This is like looking through a slit in a door frame - you move the light sensor perpendicular to the slit to see the objects inside (except not a perfect analogy: it would be more like carrying the doorframe with you as you move so that you need to move x distance to see something x distance from what you were first seeing). The image is being taken orthographically along one direction. So for the red sensor to see what the green sensor is currently seeing, the satellite must move forward by a few inches so that the red sensor is above the same spot as the green sensor was earlier. This is not how traditional photography works. There is actually a one-to-one correspondence of motion of the sensor with displacement of the imaged object (in one direction). This is similar to how a "photo finish" is done [1] Since the satellite moves much faster than the plane and the displacement between the sensors is small, it seems that the plane should not have traveled very far. [1] http://en.wikipedia.org/wiki/Photo_finish http://en.wikipedia.org/wiki/Photo_finish
- mannykannot 13y ago
  You are assuming that the lines of sight of each of the arrays are parallel with one another, so that their separation on the ground is the same as the separation of the arrays on the satellite. If they are at an angle to one another, the separation of these lines at the surface will be different from (and possibly in a different order than) the separation of the arrays. It is the surface separation, divided by the speed of the satellite, that determines the time between images of the same point on the surface by different arrays. If the arrays all share the same lens, that would cause this effect.
  - tgb 13y ago
    I'm not sure what you're saying. I think you are either saying the same thing as InclinedPlane (see my response) or are suggesting that either intentionally or due to tolerances in machining or some such, the red and green sensor units are at some small angle from each other. Assuming you're saying the latter, since the distance is large, this small angle contributes a large displacement on the ground. Afterwards the image would be stitched together to compensate for that via some calibration routine which would work for only stationary objects. That is entirely plausible, I hadn't considered that before and had just been assuming that the satellite construction was 'ideal'. This sounds like the most likely reason for such an artifact to occur, though I still wonder whether this particular imperfection would be done given the high level of precision necessary for the entire project.
    - InclinedPlane 13y ago
      It's not due to imprecision, it's due to design and fundamental optical constraints. There's nothing special about the optics used for a push broom imager, it's just an ordinary lens. If you put a full frame sensor at the focal plane you could take an image of a several kilometer square patch of Earth. If instead you put several different lines of sensors in the focal plane and put each behind a color filter then you'd have a multi-spectral push broom sensor. But each line does not view the same part of the Earth at the same time. Because the sensors scan the Earth they end up imaging the same spot within a fairly short delay. But optically it's not possible to have each color line viewing the same horizontal slice of the image of Earth at the same time, that's not the way lenses work.