12
\$\begingroup\$

I came across this collection of sample satellite images with very large files - for instance, one photo has 11000 x 9000 pixels and its file size is 49 MB.

However, what I find strange is that they are listed as:

  • Resolution: 50 cm,
  • Resolution: 65 cm,
  • Resolution: 1 m,
  • Resolution: 1.5 m, and
  • Resolution: 5 m

What that does even mean?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ Resolution in this context usually approximates pixel size. What does weight (in MB) mean I am less certain... \$\endgroup\$ Sep 26, 2017 at 19:36

2 Answers 2

21
\$\begingroup\$

Spatial resolution in Remote Sensing is normally expressed in distance per sample rather than samples per unit distance. The same information is there, it's just expressed as the reciprocal of what is usual in normal graphics fields. The "per sample/pixel/whatever" is normally dropped as well and as it is considered implicit.

So a resolution of 0.5 m is really 0.5 m/px, which is 2 px/m or 0.0508 DPI.

This means that the pixels on the ground are squares that size (at least on average, under ideal conditions) A 0.5 m resolution raster data set has pixels that are 0.5 m squares of the Earth's surface. This gets more complicated once you start considering projection of the curved surface of the earth into a flat raster grid.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Interesting. I had never heard of satellite image resolution expressed in DPI. \$\endgroup\$
    – Michael
    Sep 27, 2017 at 1:41
  • 2
    \$\begingroup\$ Presumably it's because most people really don't like fractions so for situations where the pixels are smaller than our typical units we use pixels per unit, and then for situations where they are larger than those units, we use units per pixel. Practically speaking though, there's no difference just like switching between period and frequency. \$\endgroup\$
    – smithkm
    Sep 27, 2017 at 10:18
13
\$\begingroup\$

"Resolution: <whatever>" means that the image contains sufficient detail to resolve an object of size <whatever> on the ground. There are multiple definitions of exactly what it means to be able to "resolve an object" of a given size, but basically it means that you will be able to distinguish objects that far apart, but not if they're closer. For example, "Resolution: 1 m" means you'd be able to distinguish objects which were 1 m apart, but not objects which were 50 cm apart.

\$\endgroup\$
4
  • 4
    \$\begingroup\$ Philip is spot on -- consider two automobiles are parked with a space of 2 meters between. If the resolution is 5 meters, then the two vehicles will appear as 1. If the resolution is 1.5 meters, the image will reveal 2 vehicles with space between. \$\endgroup\$ Sep 26, 2017 at 20:46
  • 4
    \$\begingroup\$ Note that this interpretation differs (slightly, probably be a factor of 2) from the one in the other answer. If you have one pixel per (square) meter, you will not be able to distinguish two objects which are 1m apart from one bigger object. \$\endgroup\$
    – Carsten S
    Sep 27, 2017 at 9:41
  • \$\begingroup\$ @CarstenS The Nyquist limit also holds for signals sampled in space rather than time, so you can indeed only distinguish one object every two meters if your resolution is one meter. \$\endgroup\$
    – Sanchises
    Sep 27, 2017 at 14:32
  • \$\begingroup\$ This is more of a Rayleigh Criterion resolution. Unfortunately, the resolution of a satellite being dependent on turbulence, ground motion, etc., this can only be expressed as 'under ideal conditions,' whereas the purely geometric "resolution" of ground grid to pixel is at least invariant. \$\endgroup\$ Sep 27, 2017 at 18:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.