Determining the height a photo was taken from

[480sparky]

I have an old photo (taken in 1909) that I have determined precisely where (relative to the ground) it was taken from, but I'm wondering if it's possible to use the features in the image to mathematically calculate the height at which it was taken from.

I have positively identified 6 houses that are still standing (easy to do with Google street view), so the latitude and longitude are easily determined. But I'm wondering if there's a way, using the know distances between these houses, to figure out how high the camera was when it took the image.

Here is the image:

The houses with the red dots are still standing. The image was taken from above this location.

 

[tirediron]

If the streets are still there and relatively unchanged, just send your drone up vertically at that distance 'til you get the same view. I get that it's not a mathematical solution, but I don't think there's a practical mathematical way to do this with any accuracy.

 

[480sparky]

If the streets are still there and relatively unchanged, just send your drone up vertically at that distance 'til you get the same view. I get that it's not a mathematical solution, but I don't think there's a practical mathematical way to do this with any accuracy.

Well, two issues arise. One is the drone won't have the same vertical FOV as this image. Second, the drone doesn't take panoramic views. I'd have to send it up, take a series of images, land, offload the images, stitch a pano, compare it to the original, and.............. do it all over again.

I'm not asking for a 1/16" tolerance here. 20 or 30 feet would be close enough.

 

[tirediron]

I never said it was an easy method!

 

[480sparky]

Nothing in my life ever is............

 

[Trever1t]

yes there is. Known height of structure. Scale. Math. Beyond me but like that.

 

[tirediron]

yes there is. Known height of structure. Scale. Math. Beyond me but like that.

That will allow you to calculate distance between two objects, but in order to determine elevation, you need to account for slant range, which would mean you would need to know the FoV of the lens at the very least, wouldn't it?

 

[480sparky]

If you need a known FOV, just crop the image so it covers 90°. Easy to do with the streets plainly visible.

 

[jake337]

Can you see what you are shooting from the ground while the drone is up there?

 

[petrochemist]

I suspect it can be approximated by comparing the apparent distance apart of two known points traveling directly away from the viewpoint with the apparent distance for points perpendicular to that (equidistant to the camera).
Points at the same distance from the camera will not have their apparent distance effected by the height, while objects directly in line with a spot directly under the camera will look to be further apart as the viewpoint gets higher.

If the streets are unchanged you should have enough points so make some reasonable estimates...

Alternatively you could ask pilots at the local airport for their estimate. They'd probably be fairly good at estimating height from the view

 

[480sparky]

Can you see what you are shooting from the ground while the drone is up there?

Absolutely. But there's no way to compare it to the original image.

 

[DScience]

If the streets are still there and relatively unchanged, just send your drone up vertically at that distance 'til you get the same view. I get that it's not a mathematical solution, but I don't think there's a practical mathematical way to do this with any accuracy.

In other words, you have no real idea whether the OP's question is possible. Man these forums are useless sometimes!

 

[DScience]

Wouldn't you have to also know the exact details of the older photo? i.e. lens focal length?

 

[tirediron]

If the streets are still there and relatively unchanged, just send your drone up vertically at that distance 'til you get the same view. I get that it's not a mathematical solution, but I don't think there's a practical mathematical way to do this with any accuracy.

In other words, you have no real idea whether the OP's question is possible. Man these forums are useless sometimes!

Get up on the wrong side of the cave this morning did we? I'm pretty sure I stated that I don't think it's possible with any accuracy. Another way to state that is, "I believe that this can be done, but the result won't be accurate".

 

[Dave442]

I think you could make some calculations by trying some different focal lengths that might have been popular for such a photo.

You can calculate the scale based on the distances in the photo and the actual distances. Then you can use that scale and using an assumed focal length (and probably need to figure in the negative size and if any enlargement was done when making the prints) and can then solve for the altitude.

Here is a page with some sample calculations...
7

This is another article that uses the known focal length/sensor size of the GoPro to do similar calculation of altitude...
How to calculate altitude from a aerial photography taken with a GoPro3be.

I guess it depends a bit on just how accurate you need to be in this calculation. Sounds like a fun project.