Process images from different heights

[alex.s]

Hello to everyone,

I'm a beginner in the field of photogrammetry so some of my questions may be misplaced.
The main idea of my question is to understand what happens when processing imagery from an unorthodox flight pattern.
I want to map an area containing a hill with relatively large height variations, between the foot and the top of the hill, compared to the flight height of the UAV.
The normal way would be to fly at a constant altitude above the top of the hill.
But that presents the problem that the ground sampling distance would vary between low and high regions of the map.
So I was thinking about two alternatives:
a) Capture nadir images from a plane of flight parallel to the surface of the hill so as to minimize the variation of distances between the camera and the ground.
b) The same as (a) but with oblique photographs having a line of sight perpendicular to the surface of the hill so as to minimize even further the variation of distances.
c) Divide the area in subregions photographed from different heights, each of which would be covered with nadir images taken from the same height, and then merge the processed chunks so the variation inside each chunk would be smaller than the total variation.
Is any of the above even possible?
Theoretically the second approach doesn't differ mathematically from the case of capturing photos of a flat ground surface the "normal" way.
I'm planning to place and measure with survey grade equipment quite a few GCPs.

Thanks
Alex

[bigben]

https://m.youtube.com/watch?v=NuZUSe87miY flight path to approximate a constant altitude

[alex.s]

Thank you very much for the reply.
I think this video depicts the first scenario.
So as I understand things this is not an uncommon practice in photogrammetry.
If this is so, can Agisoft process these sets?
I've read in http://www.agisoft.ru/forum/index.php?topic=1707.msg8940#msg8940 that it is possible but depends on more factors. Would these factors be the GCPs?
I'm going to do a little bit of testing myself over the next few days to compare the accuracy of the "normal" way compared to the (a) scenario and I'll try to post back here the results.
It's just that I don't know the specifics of the algorithms employed in each software package so although I understand that it would be possible I can't be sure for the result.
Thanks again for your help.

[bigben]

As I don't have a UAV I've been using screen grabs from 3D applications for testing various shooting scenarios. These include Google Earth (cropping out all interface elements, and including 3D buildings) and Skyrim (aerial, interior, mines, buildings etc...), although that doesn't make use of GPS positioning of the cameras. The good thing is that it's a quick way of testing different scenarios without having to go out and reshoot 

[stihl]

Our UAV system constrains it's altitude by the ground elevation. So if the ground goes up, the uav goes up accordingly. If the ground goes down, the uav goes down accordingly. That way we keep our GSD the same over the entire dataset.