Re: OpenDroneMap-dev Digest, Vol 2, Issue 8

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Re: OpenDroneMap-dev Digest, Vol 2, Issue 8

Cole
You don't want a video, you want still images. I can upload some later.
The problem with the video is that it's much lower resolution than time
lapse shots from the same camera and you end up grabbing frames from it
anyways.


1.  A series of images will be fine, 20 would images would suffice.  It will allow me to test the calibration againt a known quantity -- the goPro camera.  Once I have these I will write a test case.

2. From my understanding point cloud reconstruction does not always benefit from higher resolution images and adds quite a bit of computing overhead.  I am working with a fellow student to write up a grant proposal for real-time dense point cloud reconstruction from a MAV.  I am starting some work with LSD-SLAM, and will hopefully have some point clounds we can test.  It will be interesting to see if there is a quality trade off using the LSD-SLAM method.  
 
Measuring the checkerboard loss from really close up shots is exactly
how I guessed the new sensor size.


Where do I need to define the corrected sensor size if crop, EXIF?
 



v/r
NJK



_______________________________________________
OpenDroneMap-dev mailing list
[hidden email]
http://lists.osgeo.org/cgi-bin/mailman/listinfo/opendronemap-dev
Reply | Threaded
Open this post in threaded view
|

Re: OpenDroneMap-dev Digest, Vol 2, Issue 8

Alex Mandel-2
On 03/19/2015 09:07 PM, Cole wrote:

>>
>> You don't want a video, you want still images. I can upload some later.
>> The problem with the video is that it's much lower resolution than time
>> lapse shots from the same camera and you end up grabbing frames from it
>> anyways.
>>
>>
> 1.  A series of images will be fine, 20 would images would suffice.  It
> will allow me to test the calibration againt a known quantity -- the goPro
> camera.  Once I have these I will write a test case.
>
> 2. From my understanding point cloud reconstruction does not always benefit
> from higher resolution images and adds quite a bit of computing overhead.
> I am working with a fellow student to write up a grant proposal for
> real-time dense point cloud reconstruction from a MAV.  I am starting some
> work with LSD-SLAM, and will hopefully have some point clounds we can
> test.  It will be interesting to see if there is a quality trade off using
> the LSD-SLAM method.
>

Point cloud construction maybe, but orthophotos those pixels are data
for remote sensing purposes. Detecting type of building, plant, etc
depends on the spectral signature.


>
>> Measuring the checkerboard loss from really close up shots is exactly
>> how I guessed the new sensor size.
>
>
>
> Where do I need to define the corrected sensor size if crop, EXIF?
>

I haven't found a spot in the EXIF for the sensor size. So I've taken to
changing the Camera Model by adding a U. That way we can list it
separately in the ccd_defs.json file.

Uploading sample photos now. Will send url when they are up.

-Alex
_______________________________________________
OpenDroneMap-dev mailing list
[hidden email]
http://lists.osgeo.org/cgi-bin/mailman/listinfo/opendronemap-dev