How to give spatial information to a digital photograph?

3205
2
12-22-2014 02:17 PM
RickyStevens
New Contributor

Hi GIS community,

 

My name is Ricky. I research iceberg calving events in the Arctic. Iceberg calving defines the sudden breaking away of ice chunks from the cliff face of glaciers that move into the ocean. I use time-lapse cameras to film the calving process as the cameras allow me to get within 300 meters of the calving action without imminent death from falling ice pieces that can be the size of an average house or 12 story office block! Right now I have a couple of thousand digital photographs from which I need to accurately map the location and size of each detaching iceberg. The issue I have is that the digital photographs have no spatial information and also contain perspective errors. I have only a handheld GPS co-ordinate for the camera that took the photos.   see attached photo.@

 

For me - the location of the event needs to be in x, y, and z coordinates while the size of the event ideally needs to be in area (which I combine with an ordinal category of extra small (XS), S, M, and L.  

 

I will also soon have a surface DEM for the same location of the ice cliff at which the digital photographs were recording.

 

I was told to initially use Adobe Photoshop to draw polygons around each individual iceberg, and then to determine the centroid pixel co-ordinates of that polygon / iceberg, and then to use the subsequent photo pixel coordinates as a means to locate the same location (iceberg event) on the surface DEM.

 

So my first question is this: Are digital photograph pixel coordinates of any use for identifying the same location on a DEM? Would the DEM recognize the photo pixel coordinates?

 

Second question linked to the above: Could I transfer my polygons made in Photoshop to the DEM as a means to find the 3D area of the iceberg so I can get a handle on the iceberg size?

 

My thinking:

 

Which leads to a third and fourth set of questions: Is it possible to warp one of my digital photographs onto the DEM? This would give my photo a spatial reference. I then wanted to outline the polygons onto the warped photo so that my area calculation was in 3D and devoid of perspective issues. Is that possible?

 

Finally - if I can warp one of the photos, how would I provide spatial information to the rest of my photo batch? Would I need to manually warp every single photo on the single DEM I will soon have?  Or is there a way to warp just one image and use the meta data to link this to the rest of the photo batch? My photo batch is focused on the same section of the cliff face - so essentially the space being filmed doesn't change - it's always the same section of the cliff face in view. The camera did move about in times of strong wind so there are occasions where a little bit more of the ice cliff comes into view.

 

Thanks everyone - I basically want the most time efficient and robust way of giving real world coordinates to iceberg calving events from thousands of 2D photos that have no spatial information / coordinates. And I have a DEM (cm scale resolution) which I am hoping can help me with this!

 

Merry Christmas and any advice would be very much appreciated,

 

Ricky

0 Kudos
2 Replies
GregoryElwood1
New Contributor III

Wow! After looking at your attached photograph, all I can say is: "what a nightmare!"

Out of chaos, however, often comes order.

Maybe get a laser rangefinder and get some distance readings that you could use to estimate the size and movement of the ice blocks in your photos.

Play around with creating ice block footprints. Bring these in to ArcScene and extrude them. Maybe play around (a lot!) with 3D editing to get your extruded shapes to look more realistic. Possibly integrate the 3D ice blocks with whatever DEM or other elevation/surface data you may have. Try draping some of the photos over the 3D geometry that you create.

I don't know that I would use Photoshop for this, but that's just me. You might be able to get something really good from PS.

Between geotagging the photos and laser rangefinding the ice blocks, perhaps you can get some imagery that can be applied to modeling the events, rather than just having photographs that provide a visual reference.

Maybe. Perhaps. I've never done anything like this, so feel free to toss my suggestions.

Stay warm and good luck!

Gregory

0 Kudos
MarkBockenhauer
Esri Regular Contributor

Ricky,

One approach would be to geo-reference the photos using ArcMap and your DEM data.   Geo-reference the photo manually by clicking position on photo and then clicking the location on the DEM where it should align.   It won't be perfect, but you may be able to create something that is suitable.   Since all of the photos are taken from the same location of the same location, you can use a file copy and rename to apply the same geo-referencing information (files) to all of the other photos.  The .JGw and .aux.xml files just need to be renamed to match your other photos.

gereffiles.png

Here is a quick and simple example using ArcGIS Explorer Desktop.. I don't know where your photo is supposed to be, I just picked a glacier at random to geo-reference too.  It's flat on the surface, as I don't have a DEM.

georeferencetest.png

This doesn't help too much in the way of coming up with a size measurement for ice bergs though.  Your DEM is really the key to this.   Do you have a picture that is at the same time as the DEM data?  Preferably a picture of and DEM of a "calf".   Measure the DEM to determine the size of the iceberg and correlate that to what you see in the picture.  You may be able to come up with a good estimate for iceberg size by analyzing the photos of the calving event for that position.