I have been trying to figure out for a while now how best to orthorectify historic aerial photographs from around 1940. These photos have fiducials but no camera calibration files. I read on several sites that there are only a couple of open source solutions to achieve this, one of them being GRASS. Over the last few weeks I have been learning how to use GRASS. I have discovered that i.ortho.photo does not work for Windows so I am using GIS-Knoppix. I am beginning to understand how to set up a project in GRASS (on top of learning Linux).
I have a few questions/comments:
- When is i.ortho.photo going to be available for the Windows binaries?
- The historic aerial photos do not have camera calibration files. I have searched all over and they do not seem to exist anymore. Is there a way to orthorectify the images without the camera calibration files? Can I make some assumptions? If so, what parameters should I enter?
- When I tried using photo.2image, I was not prompted with “Look ok? (Y/N)” so I was not able to mark the fiducials.
I would appreciate any help! Thanks!
Shaun
Shaun & Melissa Busler pisze:
1) When is i.ortho.photo going to be available for the Windows binaries?
Dunno. It currently depend on X displays, not easily portable to else than Unix.
2) The historic aerial photos do not have camera calibration files. I have
searched all over and they do not seem to exist anymore. Is there a way to
orthorectify the images without the camera calibration files? Can I make
some assumptions? If so, what parameters should I enter?
If the DPI at which photos have been scanned is known to you, you can measure the principal point and fiducial marks coordinates in any image editing software, like GIMP. Interior orientation results will depend on how close is the geometry of the image plane to the geometry of camera's optical system, and if/how the photo scan is distorted.
Idea, not tested: you might try using software like MapAnalyst [1] to asses the distortion of the image, which can help you guessing genuine fiducial marks coordinates.
Then you need the focal length. It often is present on the photo border. Or guess it :).
3) When I tried using photo.2image, I was not prompted with "Look ok?
(Y/N)" so I was not able to mark the fiducials.
photo.2image is not supposed to be used standalone. i.ortho.photo is the UI for photo.2image & friends.
There's plenty of info you might find usefull on the net. eg. [2] [3].
[1]http://mapanalyst.cartography.ch
[2]http://support.erdas.com/focus3/files/Aerial_photo.pdf
[3]http://www.pcigeomatics.com/services/support_center/faqs/oe_general_v10.html#no_fiducial
Thank you so much for your response! The resources you linked to were very useful. The Erdas paper is especially good. I never heard of MapAnalyst. I will have to give it a try.
Unfortunately, the focal length of the camera is not on the photo border. The photo must be too old. Do you have any recommendations on what focal length I should try? It was taken in 1939 for the USDA Agricultural Adjustment Administration. If it would help, I could send the image file. It is about 7 mb in size.
I was using photo2.image as part of the i.ortho.photo UI when I was having the problems. It may be because I am using GIS-Knoppix. If I use GIMP to measure the fiducial mark coordinates, do I still need to use photo2.image or do I just use photo.camera?
Thanks!
Shaun
On Sun, Apr 6, 2008 at 6:18 AM, Maciej Sieczka tutey@o2.pl wrote:
Shaun & Melissa Busler pisze:
- When is i.ortho.photo going to be available for the Windows binaries?
Dunno. It currently depend on X displays, not easily portable to else than Unix.
- The historic aerial photos do not have camera calibration files. I have
searched all over and they do not seem to exist anymore. Is there a way to
orthorectify the images without the camera calibration files? Can I make
some assumptions? If so, what parameters should I enter?
If the DPI at which photos have been scanned is known to you, you can measure the principal point and fiducial marks coordinates in any image editing software, like GIMP. Interior orientation results will depend on how close is the geometry of the image plane to the geometry of camera’s optical system, and if/how the photo scan is distorted.
Idea, not tested: you might try using software like MapAnalyst [1] to asses the distortion of the image, which can help you guessing genuine fiducial marks coordinates.
Then you need the focal length. It often is present on the photo border. Or guess it :).
- When I tried using photo.2image, I was not prompted with “Look ok?
(Y/N)” so I was not able to mark the fiducials.
photo.2image is not supposed to be used standalone. i.ortho.photo is the UI for photo.2image & friends.
There’s plenty of info you might find usefull on the net. eg. [2] [3].
[1]http://mapanalyst.cartography.ch
[2]http://support.erdas.com/focus3/files/Aerial_photo.pdf
[3]http://www.pcigeomatics.com/services/support_center/faqs/oe_general_v10.html#no_fiducial
On Thu, Apr 10, 2008 at 5:29 PM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
Unfortunately, the focal length of the camera is not on the photo border.
The photo must be too old. Do you have any recommendations on what focal
length I should try? It was taken in 1939 for the USDA Agricultural
Adjustment Administration.
You could try 150mm or 305mm. In fact, i.ortho.photo is quite robust,
just try and hope that it converges.
(and/or: Maybe someone on the PROJ4 list knows.)
Markus
May it be a stupid suggestion, but did you try to contact anyone at the
USGS ? perhaps they maintain data from these old times... (here in
France IGN is always helpful, and can provide answers to such historical
requests)
Vincent.
Le jeudi 10 avril 2008 à 18:33 +0200, Markus Neteler a écrit :
On Thu, Apr 10, 2008 at 5:29 PM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
> Unfortunately, the focal length of the camera is not on the photo border.
> The photo must be too old. Do you have any recommendations on what focal
> length I should try? It was taken in 1939 for the USDA Agricultural
> Adjustment Administration.
You could try 150mm or 305mm. In fact, i.ortho.photo is quite robust,
just try and hope that it converges.
(and/or: Maybe someone on the PROJ4 list knows.)
Markus
_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
Would it be possible to calculate these parameters with a panoramic creation package like panotools?
http://wiki.panotools.org/Lens_correction_model
It calculates a value for field of view, but I didn’t notice an actual focal length parameter. Maybe this is easy to associate?
One benefit is that you are calculating using image overlap, which may be easier than trying to reference to a recent photo.
-Jamie
On Thu, Apr 10, 2008 at 9:53 AM, Vincent Bain <bain@toraval.fr> wrote:
May it be a stupid suggestion, but did you try to contact anyone at the
USGS ? perhaps they maintain data from these old times… (here in
France IGN is always helpful, and can provide answers to such historical
requests)
Vincent.
Le jeudi 10 avril 2008 à 18:33 +0200, Markus Neteler a écrit :
On Thu, Apr 10, 2008 at 5:29 PM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
Unfortunately, the focal length of the camera is not on the photo border.
The photo must be too old. Do you have any recommendations on what focal
length I should try? It was taken in 1939 for the USDA Agricultural
Adjustment Administration.
You could try 150mm or 305mm. In fact, i.ortho.photo is quite robust,
just try and hope that it converges.
(and/or: Maybe someone on the PROJ4 list knows.)
Markus
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
Shaun & Melissa Busler pisze:
Thank you so much for your response! The resources you linked to were very
useful. The Erdas paper is especially good. I never heard of MapAnalyst.
I will have to give it a try.
Unfortunately, the focal length of the camera is not on the photo border.
The photo must be too old. Do you have any recommendations on what focal
length I should try?
In one of the links already mentioned [1] there is a couple of hints. Maybe try asking on some RS-related ML?
I was using photo2.image as part of the i.ortho.photo UI when I was having
the problems. It may be because I am using GIS-Knoppix. If I use GIMP to
measure the fiducial mark coordinates, do I still need to use photo2.image
or do I just use photo.camera?
In GIMP measure fiducial marks coordinates, assuming you have the resolution of the image set right. In GIMP you can check it in Image > Print size - if the value there is wrong, set a correct one.
Then, in i.ortho.photo go through steps 1-3. In step 4 enter a name your camera parameters will be stored under. A dialog pops up for camera parameters. Once you enter the number of fiducial marks and press Esc+Enter, you will be prompted with a dialog for their coordinates - that's where to put values measured in GIMP.
Maciek
[1]http://www.pcigeomatics.com/services/support_center/faqs/oe_general_v10.html#no_fiducial
May it be a stupid suggestion, but did you try to contact anyone at the
USGS ? perhaps they maintain data from these old times... (here in
France IGN is always helpful, and can provide answers to such historical
requests)
Vincent.
Similarly in Canada, provincial Department of Natural Resources frequently
maintain documents from the original flight transects, sometimes going back
several decades. I've managed to recover much aerial photo detail by contacting
these agencies, including fiducial mark coordinates, camera make and focal length,
aircraft altitude, lens type, etc.
It's probably just a matter of tracking down the right federal agency to contact.
~ Eric.
I wish the same was the case for the US for this period of aerial photos. I have tracked down the location of the photos to the National Archives and they do not have the calibration reports. Prior to the archives the USDA was responsible for the photos; however, they sent “everything” they had to the Archives. The USGS could provide me a calibration report if I knew the type of camera or serial number of the camera lens, but I do not have that either. I even tried talking to the contractor who took the photos since they are still around, but they do not have any files prior to the 1970’s.
The records get better after the early 1940’s. For some reason, they do not have any information for these early aerial photos.
Thanks!
Shaun
On Thu, Apr 10, 2008 at 2:46 PM, Patton, Eric <epatton@nrcan.gc.ca> wrote:
May it be a stupid suggestion, but did you try to contact anyone at the
USGS ? perhaps they maintain data from these old times… (here in
France IGN is always helpful, and can provide answers to such historical
requests)
Vincent.
Similarly in Canada, provincial Department of Natural Resources frequently
maintain documents from the original flight transects, sometimes going back
several decades. I’ve managed to recover much aerial photo detail by contacting
these agencies, including fiducial mark coordinates, camera make and focal length,
aircraft altitude, lens type, etc.
It’s probably just a matter of tracking down the right federal agency to contact.
~ Eric.
Markus,
Thanks for the suggestion. I will give those focal lengths a try.
What do you mean by i.ortho.photo is robust and hope that it converges?
Shaun
On Thu, Apr 10, 2008 at 12:33 PM, Markus Neteler <neteler@osgeo.org> wrote:
On Thu, Apr 10, 2008 at 5:29 PM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
Unfortunately, the focal length of the camera is not on the photo border.
The photo must be too old. Do you have any recommendations on what focal
length I should try? It was taken in 1939 for the USDA Agricultural
Adjustment Administration.
You could try 150mm or 305mm. In fact, i.ortho.photo is quite robust,
just try and hope that it converges.
(and/or: Maybe someone on the PROJ4 list knows.)
Markus
Jamie,
I am not familiar with panotools. I am willing to give it a try though!
Shaun
On Thu, Apr 10, 2008 at 1:16 PM, Jamie Adams <dickeya@gmail.com> wrote:
Would it be possible to calculate these parameters with a panoramic creation package like panotools?
http://wiki.panotools.org/Lens_correction_model
It calculates a value for field of view, but I didn’t notice an actual focal length parameter. Maybe this is easy to associate?
One benefit is that you are calculating using image overlap, which may be easier than trying to reference to a recent photo.
-Jamie
On Thu, Apr 10, 2008 at 9:53 AM, Vincent Bain <bain@toraval.fr> wrote:
May it be a stupid suggestion, but did you try to contact anyone at the
USGS ? perhaps they maintain data from these old times… (here in
France IGN is always helpful, and can provide answers to such historical
requests)
Vincent.
Le jeudi 10 avril 2008 à 18:33 +0200, Markus Neteler a écrit :
On Thu, Apr 10, 2008 at 5:29 PM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
Unfortunately, the focal length of the camera is not on the photo border.
The photo must be too old. Do you have any recommendations on what focal
length I should try? It was taken in 1939 for the USDA Agricultural
Adjustment Administration.
You could try 150mm or 305mm. In fact, i.ortho.photo is quite robust,
just try and hope that it converges.
(and/or: Maybe someone on the PROJ4 list knows.)
Markus
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
On Fri, Apr 11, 2008 at 5:09 AM, Shaun & Melissa Busler
<buslers@gmail.com> wrote:
Markus,
Thanks for the suggestion. I will give those focal lengths a try.
What do you mean by i.ortho.photo is robust and hope that it converges?
It won't explode 
I mean that even if you are off from the true value it
may still get out a result.
Markus
PS: The algorithms is described in
Elements of Photogrammetry, With Air Photo Interpretation and
Remote Sensing
by Paul R. Wolf, 562 pages
Publisher: McGraw Hill Text; 2nd edition (January 1983)
(get that for cheap as second hand book. if interested..)
Hi all,
- Very old aerialphotographs for which you do not have supporting data like focal lengths etc, except a stereo pair of photos, how to convert them to digital ??
- Can a normal (Hp) A4 scanner work for this.
- Any one on GRASS list tried it this way.
Ravi Kumar
Markus Neteler neteler@osgeo.org wrote:
On Fri, Apr 11, 2008 at 5:09 AM, Shaun & Melissa Busler
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com