Remote Sensing Lab 7

Goal and Background

The goal of this lab is to gain photogrammetric skills for use in correcting aerial photographs and satellite images. Specific skills include calculating scale, measuring perimeter and area of features, calculating relief displacement, and performing orthorectification on satellite images.

Photogrammetry itself is the science of making accurate measurements from aerial photographs and satellite images to create accurate products from such photos and images. If the image is not corrected for errors such as perspective convergence, scale variation, or relief displacement, then the measurements and subsequent products will not be geometrically accurate. Thus, photogrammetry skills are a necessary tool in remote sensing and are covered in this lab to introduce such skills.

Methods

Part 1: Scales, measurements, and relief displacement

Two methods for scale calculation were used. The first involved dividing the photo distance by the real-world distance.

S = Photo distance/real-world distance

 The second involved dividing the focal length of the camera by the height of the camera minus the height of the object. Each scale calculation method has its merits, but the latter is easier because ground work is not needed.

S = f/H-h

The measurements of areas of features on aerial photographs was completed using the ‘measure perimeters and areas’ digitizing tools in Erdas Imagine.

*polygon tool*

Relief displacement is the displacement of tall objects and features on an aerial photograph from their true planimetric location on the ground. To calculate relief displacement (d), the height of the object in the real world was multiplied by the distance of the center of the object from the principle point and then divided by the height of the camera above the datum.

d = h x r/H

The height of the object in the real world was calculated by measuring the photograph and then multiplying it by the scale. The distance r was measured on the photograph as well. The height of the camera was given.

Part 2: Stereoscopy

In this section GCPs were used to show a 3D view, or anaglyph, of the City of Eau Claire, WI. First, an image of the City of Eau Claire was brought into one viewer on Erdas. Then, a DEM of the same area was inputted to another viewer. Using the anaglyph tool in Erdas, the anaglyph image was generated. This same process was repeated except with a LiDAR-derived DSM of the City of Eau Claire. The DSM-derived anaglyph was then compared to the DEM-derived anaglyph.

Part 3: Orthorectification

In this section a planimetrically correct orthoimage was created using the Erdas Lecia Photogrammetric Suite (LPS). A SPOT satellite image and an orthorectified aerial photo were both used as the sources for ground control measurements.

First, a new project was created using the LPS project manager. The project geometric model was a polynomial-based pushbroom using the SPOT pushbroom. The Horizontal coordinate system was set to UTM Clark 1866 with the NAD27 (CONUS) datum zone 11 north. The vertical datum was set to WGS 1984. After inputting the SPOT satellite image, the sensor’s information was verified. In the point measurement tool, the GCP reference source was set to image layer because an image that has already been orthorectified will be used to collect horizontal control points. Two ground control points were manually inputted using an orthorectified reference image. 

After the first two points were collected, the automatic (x.y) drive function was used to approximate the GCP location in the image file based on the GCP position in the reference image. For GCP three the location still had to be inputted because the function did not accurately locate the GCP location on the SPOT image. GCPs 3-9 were collected in this manner. 

For GCPs 11 and 12, an orthoimage with a different horizontal reference source was used. This image had a scale of  1:40,000 with a spatial resolution of 2 meters (the point label 10 was skipped to show the boundary between the two horizontal reference sources). 

Next, elevation information was updated for all the horizontal reference GCPs previously obtained using a DEM. Then the type and usage for each control point was specified. The type was specified as full, meaning there were x, y, and z coordinates for each GCP. The usage for each GCP was set to ‘control.’ 

The same process above was repeated for the second image except the GCPs in the second image were collected based on those already collected for the first image (figure 1). Some GCPs in the first image were not in the second image, so they were not collected. 

Figure 1. GCP collection for the second image using GCPs collected in the first image.

The next step was to perform the tie point collection process which measures the image coordinate positions of ground points appearing on the overlapping area of the two SPOT images. This is automated by LPS (figure 2).

Figure 2. Tie point collection process in Erdas.

Following tie point collection, triangulation must be performed. This establishes the mathematical relationship between the images in the block file, the sensor model, and the ground. The iterations with relaxation was set to 3 and the report units to pixels. The ground point type and standard deviations were set to ‘same weighted values’ and the value was set to 15 for x, y, and z because the spatial resolution of one of the SPOT images was 20 meters, so the value of 15 makes sure that the GCPs are at least accurate to 15 meters. After running the model the triangulation report was verified to make sure the model ran correctly. This step supplies the exterior orientation information needed (figure 3).

Figure 3. Triangulation Report generated by Erdas.

Finally, the orthorectification process can be completed that removes relief displacements and other geometric errors to create an image that displays objects in their correct x, y positions.

After the orthorectification process is finished, the orthorectified images were viewed for the accuracy of spatial overlap at the boundaries between the two images.

Results

Part 1

Scale using photo distance: 1:39,212

Scale using camera focal length: 0.00792

Relief Displacement: 0.312 inches

 Part 2

Figure 4. a) Anaglyph image created using a DEM. b) Anaglyph image created using a LiDAR-derived DSM.

The DSM anaglyph has a better output compared to the DEM anaglyph. In the DSM anaglyph, terrain features such as trees and small homes appear in 3D whereas in the the DEM anaglyph these features appear flat and pixelated. The edges of elevation changes are also sharper in the DSM anaglyph compared to the DEM anaglyph. There are a couple of reasons for the improved quality in the DSM anaglyph. One, the DSM, or digital surface model, takes into account all features on the surface, not just the ground level. Two, the DSM has a 2 meter resolution while the DEM has a 10 meter resolution. 

Part 3

Figure 5. Final orthorectified images created using Erdas Lecia Photogrammetric Suite (LPS).

The two orthorectified images have a high degree of spatial overlap at the boundaries. Upon first inspection, ridges and fluvial features that have linear aspects transition seamlessly into each other. When the swipe function was used, the fluvial features still lined up. In areas with roads or other man-made (right angle) features, the lines transition into each other very nicely as well. The swipe function confirms the spatial accuracy at the boundary of overlap for man-made features (figure 6).

Figure 6. Zoomed in view of the area of overlap between the two orthorectified images.

Sources

National Agriculture Imagery Program (NAIP) images are from United States Department of Agriculture, 2005.
Digital Elevation Model (DEM) for Eau Claire, WI is from United States Department of Agriculture Natural Resources Conservation Service, 2010.
Lidar-derived surface model (DSM) for sections of Eau Claire and Chippewa County are from Eau Claire County and Chippewa County governments.
Spot satellite images are from Erdas Imagine, 2009.
Digital elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009. 
National Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009.