Forum

Author Topic: Images acquired from high distance  (Read 10201 times)

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Images acquired from high distance
« on: June 25, 2016, 03:00:31 PM »
Hello,
I have a problem with images acquired from high distance.
I use a Nikon Coolpix P530 mounted on tripod with a constant focal length of 180 mm.
I photograph a rockmass from a distance of 1Km (more or less).

The problem start when I try to georefernce the point clud: I have high error (2.6 m with 13 markers). The error increase if I use much more markers.
Markers are measured with a Total Station.

My question is, if Agisoft Photoscan is suitable in order to process images taken from  high distance (1Km).

Thank you so much!

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: Images acquired from high distance
« Reply #1 on: June 25, 2016, 03:19:18 PM »
Hello giannim,

And what is the distance between the camera positions and effective pixel resolution of the images (how accuractely you can detect the markers on the images)?
Best regards,
Alexey Pasumansky,
Agisoft LLC

Paulo

  • Hero Member
  • *****
  • Posts: 1303
    • View Profile
Re: Images acquired from high distance
« Reply #2 on: June 25, 2016, 05:46:42 PM »
Hi,

as Alexey said, the distance between succesive shots is critical in the precision of depth (because of famous Base to Height ratio).

Given your camera and focal length, I would think a distance between shots of some 10 m should give a depth accuracy better than 50 cm...
Best Regards,
Paul Pelletier,
Surveyor

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Images acquired from high distance
« Reply #3 on: June 25, 2016, 06:55:55 PM »
Hi,

maybe a small base can be the problem.

The distance between two shots is 10-15 m

GSD is 7.47 mm

Paulo

  • Hero Member
  • *****
  • Posts: 1303
    • View Profile
Re: Images acquired from high distance
« Reply #4 on: June 25, 2016, 08:15:56 PM »
In Theory, you should obtain better precision with this 10 to 15 m base....

Is image quality, good?

And camera focal correctly entered in Camera calibration window? (should be around 134 400 ...
Best Regards,
Paul Pelletier,
Surveyor

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Images acquired from high distance
« Reply #5 on: June 25, 2016, 08:20:51 PM »
Yes it's good.

f is 141449

Arie

  • Full Member
  • ***
  • Posts: 134
    • View Profile
Re: Images acquired from high distance
« Reply #6 on: June 26, 2016, 01:57:44 PM »
I'd think, a baseline of 10-15 m for shooting at a distance of 1km is way too small. Additionally the 180 mm on a 1/2,3" sensor equals to a 1008 mm focal length on a full frame sensor.

I'm pretty sure you will not get decent results regardless the software you use.
Cheers.

frank.stremke

  • Full Member
  • ***
  • Posts: 206
    • View Profile
Re: Images acquired from high distance
« Reply #7 on: June 27, 2016, 11:35:00 AM »
i agree the large zoom (focal legth) is not good for multi image photogrammetry also the base should be bigger i would go for 80-100 m
frank

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Images acquired from high distance
« Reply #8 on: June 27, 2016, 01:56:13 PM »
Ok, but long-range terrestrial photogrammetry needs a long focal length.

What is your idea in order to do a correct survey?

Thank you!

stihl

  • Sr. Member
  • ****
  • Posts: 410
    • View Profile
Re: Images acquired from high distance
« Reply #9 on: June 27, 2016, 03:45:47 PM »
I'd suggest a camera with a bigger (and better) sensor. That way your footprint on the ground gets larger thus you can zoom out further to reduce your focal length but still get the required overlap and image footprint so you're not flying an excessive amount of flightlines.

A 16 megapixel camera on 1km AGL sounds inadequate.

frank.stremke

  • Full Member
  • ***
  • Posts: 206
    • View Profile
Re: Images acquired from high distance
« Reply #10 on: June 27, 2016, 03:52:58 PM »
i was about to write the same a better sensor (larger and higher resolution) would be better since you could reduce the focal legth
if you go with a 24mp full frame or even a medium format with 50+ 60mp you could come down to better levels
and also i would sugest as mentioned before to increase base distance
frank

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Images acquired from high distance
« Reply #11 on: June 27, 2016, 04:17:04 PM »
Ok, thank you so mach!

I have the possibility to use a Nikon D7100 with a Nikkor 18-140 mm.

However, if I use the D7100 with (140 focal length), I have a GSD of 27.97 mm. Instead, with the coolpix p530 the GSD is 7.47 mm.

frank.stremke

  • Full Member
  • ***
  • Posts: 206
    • View Profile
Re: Images acquired from high distance
« Reply #12 on: June 28, 2016, 10:06:01 AM »
hi
well i dont know the project specifications but i would say that is still good and also its not all about GSD especialy if you work terrestrial long range you will have lots of other problems such as refraction and the accuracy is limited by these as well.
frank

giannim

  • Newbie
  • *
  • Posts: 11
    • View Profile
Re: Images acquired from high distance
« Reply #13 on: June 29, 2016, 01:16:16 PM »
Unfortunately, I need to get close to the rock slope as much as possible.
For this reason I must use a long focal length (180 mm).

It is true that a 24mp full frame is better (improve quality and footprint gets larger) but in order to study the rock mass fractures I need to get close to the rock slope as much as possible. With the nikon d7100 (24 mp 23.5x15.6) I should use a 400 mm focal length (Unfortunately, I have a 140 max).

But the real purpose of my post is to understand if photoscan is suitable in order to process images taken from  high distance.

I must photograph a rock slope from a distance of 1 Km using a tripod setup (Terrestrial survey not aerial).
Now, the base will be bigger!

My doubt is: if I use two base and for each one I photograph with different angles (in order to photograph the whole rock slope), is Photoscan able to process the acquired images?
In other words,  I have 100 images (acquired with different angles) but with only two station (for example).

Thanks
« Last Edit: June 29, 2016, 01:32:54 PM by giannim »

Alexey Pasumansky

  • Agisoft Technical Support
  • Hero Member
  • *****
  • Posts: 14813
    • View Profile
Re: Images acquired from high distance
« Reply #14 on: June 29, 2016, 03:56:05 PM »
Hello giannim,

In case you are taking the images from the same camera position, you should do it preferably with the help of calibrated nodal head, while the cameras in PhotoScan Workspace pane should be grouped to the separate camera folders that should be assigned to Camera Station class.

But note that in this case you would have the result like similar to the single stereopair, so there might be quite a lot of blind zones and the accuracy would be worse than for the multi camera station projects.
Best regards,
Alexey Pasumansky,
Agisoft LLC