Mosaicing Historic Photos with ArcGIS

1
Mosaicing Historic Photos with ArcGIS

• James Twoteeth
• GIS Analyst
• CDA Tribe GIS

2
Background

• We obtained 1947 aerial photography from the
  Forest Service in St. Maries, Idaho.
• Last summer, we scanned the 144 photos using a
  simple Office jet scanner.
• We constructed an index layer in ArcGIS and
  hyperlinked the photos to the points.
• Many People wanted a geo-registered
• image of the 1947 photos instead of just
• unregistered images that we currently had.

3
Why Mosaic Historic Photos??

• Mosaicing Historic photos can be useful by
  allowing a user to compare time periods and to
  map changes in agriculture, roads, and land
  cover.

4
Agenda

• Topics to be covered
• Pre-processing of images (decisions??)
• Geo-registration (Ortho-rectification,
  geo-correction???)
• Mosaicing images using Image Analyst

5
Orthorectification What, Why, How, Who?  
Overview Orthorectification is simply the
process of removing scale variations from a
remotely sensed image, such as an aerial
photograph, SPOT satellite image, radar image,
etc.   Benefits for Image Mosaicing An
advantage of the orthorectification process is
the fact that many ortho images can be mosaiced
together to form a seamless image map covering
large areas.
6

• Pre-processing of images (decisions??)

After receiving the requests to mosaic the
unregistered images, we realized that we didnt
have the necessary camera information or all of
the fudical marks on the photos that were needed
to orthorectify the photos.
We decided that we could do just a straight
georegistration of the photos and mosaic them
together.
Fudical Mark???
7
Since we didnt need the fudical marks, we
decided to clip out the images using MS Photo
Editor to help us better identify similar areas
during geo-registration.
Original Scanned Image Clipped Image
8
Geo-registration (Ortho-rectification,
geo-correction???)
To do a straight geo-registration of the images
we used the geo-referencing tool in ArcGIS to
warp the images and save them as grids.
Add Control Points
Layer target
Drop down menu
Open Link Table
One thing to consider when geo-registering aerial
photography is that the aerial photography
contains distortions that are not true
representations of the ground. So the most
accurate part of the photograph is the center.
9
We then began geo-registering our images to our
current data such as roads, orthoimagery, and
GCDB. As shown in the photo, similar areas are
identified in both the image and your reference
data. You can then began to add control points
in order to warp the image. On average we had
between 15 to 20 control points when warping our
images.
10
After adding a few control points, our image
begins to stretch and look acceptable. When
adding control points, you will want to keep your
RMS error down close to zero if you can. You can
look at your control points and RMS error by
bringing up the link table in ArcGIS
When satisfied with the image, we then saved our
geo-referenced image out to a ESRI grid. It took
about 80 hours to geo-reference the 144 photos.
11
Mosaicing images using Image Analyst
Dealing with sidelap and overlap We realized
that Image Analyst doesnt really deal with
overlap of photos very well.
So we wrote an AML program that makes a polygon
coverage of the center of the photo based on the
surrounding photos.
r mkbox.aml ltcentergt ltnorthgt lteastgt ltsouthgt
ltwestgt
12
144 Polygon coverages created
We then clipped the geo-referenced grids based
upon these polygon coverages.
13
Geo registered image
Clipped Image
14
We then verified the clipped images lines up
accordingly to mosaic them.
15
(No Transcript)
16
(No Transcript)
17
1947 photos
18
Current Ortho imagery
19
Thank you Any Questions????