Lab 7: Photogrammetry
Melissa Hackenmueller
Goals
and Background
The overall goal of this lab was to
develop the skills needed to perform photogrammetric tasks on aerial
photographs and satellite images. The goals were further separated into three
parts. The goal of the first part of this lab was to ensure my understanding of
photographic scales, measurement of areas and perimeters of features on an
image, and calculation of relief displacement. The second portion of this lab
was to increase my knowledge of stereoscopy and its uses. The last goal of this
lab was to perform orthorectification on satellite images.
Methodology
Part 1
The first section of part one was designed
to improve the calculation of photographic scales. The first step was to
determine the distance between A and B on an image (Eau_Claire_West-se.jpg)
which I measured as 2.7 inches (see figure below).
The
Actual distance between the two points was measured with an engineer’s chain to
be 8,822.47 feet. The following math was done and the result was a scale of 1:
39,210.98
for the image.
Actual distance:
8822.47 ft.
Photo: 2.7 inches
8822.47 ft x 12 =
105,869.6 in
2.7 in / 105,869.6
in divide
numerator and denominator by 2.7
Scale is
1:39,210.98
The
next question was also to find a photographic scale, but using the equation S= f /H –h this time. The following
calculations were done to determine the scale to be 1:38,519.
F
= 152 mm /10/100 = 0.152 m *3.28 = 0.49856 ft
H = 20,000 ft amsl
h = 796 ft
S = 0.49856 ft / (20,000 ft – 796
ft) divide numerator and denominator by
0.49856 ft
Scale is 1:38,519
The
goal of the next section of part one was to determine the area and perimeter of
a pond on an aerial image. Using the polygon measure tool, I carefully
digitized the outline of this pond in order to get an area and perimeter (Fig.
1). The digitizing of the pond yielded an area of 37.6880 hectares and a
perimeter of 4,063.42 meters. The last section of part one was calculating
relief displacement of a smoke stack in an Eau Claire image (see figure below).
Tall
objects are distorted in aerial photographs and the equation, d= h x r / H, is used to calculate
relief displacement, which is the distance the object needed to be moved
towards the principal point in order to fix the distortion. The following calculations
gave a relief displacement of 0.228 inches on the smoke stack.
H
= 3,980 ft x 12 = 47,760 in
h = 0.4 x 3,209 = 1,283.6 in
r = 8.5 in
d = 1283.6 in x 8.5 in / 47760 in =
0.228 in
Relief displacement is 0.228 inches
Part 2
The goal of part two of this lab was
to generate a three dimensional image using a digital elevation model and LiDAR
derived surface model. In order to accomplish this two images were needed. I
imputed ec_city.img and ec_dem_2.img into Erdas Imagine. I then used the
Anaglyph Terrain tool to create a stereoscopic image. The input DEM, input
image, and output fields just need to be filled in and then the defaults can be
used to create a new image. I then did the same processes to create a stereoscopic
image with a LiDAR DSM image.
Part 3
The final part of this lab is
orthorectification. Erdas Imagine Lecia Photogrammetric Suite will be used. The
tasks of this section included creating a new project, select a horizontal
reference source, collect GCPs, add a second image to the block file, collect
another set of GCPs, perform automatic tie point collection, triangulate the
images, orthorectify the images, view the orthoimages, and save the block file.
I started by opening IMAGINE Photogrammetry in the toolbox tab of Erdas
Imagine. I then choose create a new block file from the project manager that is
open. I saved the output image in my desired location, chose polynomial-based
pushbroom for geometric model, selected SPOT Pushbroom, and then hit okay to
finish the model setup. Next, the block property setup dialog opens. I chose
UTM as the projection type, Clarke 1866 as the spheroid name, NAD27(CONUS) as
the datum name, 11 as the UTM Zone, North as the field, and then hit okay to
conclude the set up. The next step is to add the imagery to the block. To do
this I chose the add frame icon from the block project tree view, making sure
that the images folder is highlighted beforehand. The image file name dialog is
now open and I added the image Spot_pan.img. I next went to the SPOT pushbroom
frame editor to ensure that the parameters that I inputted earlier were
correct, they were so I exited.
The next step is to start adding GCPs by
clicking the start point measurement tool and using the classic tool. Next, I
clicked the reset horizontal reference source icon to open the GCP reference
source dialog. Choose the image layer button then move on to the reference
image layer dialog and imput the image xs_ortho.img. I then choose use viewer
as reference so the xs_ortho image displayed on the left and the original image
on the right (see Fig. below).
To
begin adding GCPs, I clicked add button in the point measurement tools and then
the create point icon. I clicked on the road intersection desired on the
reference image (xs_ortho.img) to input my first GCP. Once, I was happy with
the first point, I added the point to the same spot on the original image
(spot_pan.img). The first GCP was successfully added, so I followed the same
procedure above for the next 8 GCPs. I then saved my work because the last two
GCPs will be added from a different reference image. To add a new reference
image click on the reset horizontal reference source icon again, this time I
chose NAPP_2m_ortho.img as my reference. I added two more GCPs with this new
reference image following the same structure as above. Now, a vertical
reference source can be used to collect elevation information, I used the
palm_springs_dem.img. I choose the reset vertical reference source icon this
time and inputted the desired reference image. Once, the image was inputting I
selected all of my Point # column and clicked on the update Z values of
selected points icon, which updated the Z values based on the
palm_springs_dem.img. The Figure below is what my X, Y, and Z reference columns
now look like.
The next section of this lab focuses
on the set up before collecting tie points. First, the type and usage columns
need to be filled out. I selected the type column and then choose formula from
the column options. Here, I typed in “Full” and then hit apply. I repeated the
steps above for the usage column, but typed “control” this time. Now that these
columns are updated, I hit save and closed the point measurement tool. Next, I
moved to the editing of the spot_panb.img by clicking the add frame icon and
adding the image. I then highlighted row #2 and clicked the frame properties
icon. The SPOT pushbroom frame editor opens, I hit okay to accept the
parameters. Next, I used the point measurement tool to located the points that
have already been collected from the first image, spot_pan, and plot them in
the second image, spot_panb. I made sure to highlight point #1, before using
the create point icon to add the points to the spot_panb image. Some of the
points were not located on spot_panb.img, but I added all the similar points
following the same procedure as above and then saved. My images now look like
Figure 2.
The next step is tie point
collection, triangulation, and ortho resampling. From the point measurement
tool palette I chose the automatic tie point generation properties icon. Then I
used all available for image used, exterior/Header/GCP for initial type, set
the image layer used for computation to 1, set the intended number of
points/image field to 40, and hit run. This automatically set points tied to
both images, my points now totaled 37. After I checked the accuracy of the
points created, I saved them. Now, that all the tie points are found I can
perform triangulation. On the IMAGINE photogrammetry project manager, I chose
the Triangulation properties. I put the iterations with relaxation value to 3,
the image coordinate units for report to pixels, the type of point to same
weighted value, the x, y, z values to 15, then ran the triangulation, and saved
it in an appropriate spot. I started the final process of ortho resampling by
clicking on the ortho resampling process icon. I selected DEM as the source,
inputted the palm_springs_dem.img, entered the output cell size to 10 for both
x and y, saved the output file in an appropriate spot, ensured that the
resampling method was bilinear interpolation, made sure to add both
spot_pan.img and spot_panb.img, and then I hit okay to start the resampling
process. Finally, I created a ortho resampled image (Figure 3).
Results
 |
| Figure 1. This image shows the points that I collected around the pond in Eau Claire using the measurement tool to get an area and a perimeter. |
 |
| Figure 2. This image shows the GCP points that I collected on the spot_pan.img and then on the spot_panb.img. |
 |
| Figure 3. This is my final image output after the process of ortho resampling on both the spot_pan.img and spot_panb.img. |
References
Digital
Elevation Model (DEM) for Eau Claire, WI is from United States Department of Agriculture
Natural Resources Conservation Service, 2010.
Digital
elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009.
Lidar-derived
surface model (DSM) for sections of Eau Claire and Chippewa are from Eau Claire
County and Chippewa County governments respectively.
National
Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009.
National
Agriculture Imagery Program (NAIP) images are from United States Department
of Agriculture, 2005.
Spot
satellite images are from Erdas Imagine, 2009.
No comments:
Post a Comment