Lab 6: Geometric Correction

Goal

This lab is intended to familiarize the students with the preprocessing technique called geometric correction. Geometric correction is typically done before the images are sent out for use, but the process is important to understand regardless. It is comprised of image-to-map and image-to-image rectification.

Part 1: Image-to-Map Rectification

Image to map rectification uses existing maps to geometrically correct warped or distorted images collected by satellites. This is done using ground control points throughout the image. The more distorted the image is, the more GCPs you will need to correct it.

In this part of the lab, we geometrically corrected a USGS TM image of the Chicago, Illinois area to a preexisting digitized map. By distributing GCPs evenly across the image, the software can stitch the new image to the map's outline. This makes the new image geometrically tied to a correct map, therefore making the new image geometrically correct. The RMS error is the root mean square difference between the GCPs on both images. The lower the RMS error, the closer the point is to its correct location.

(see figure 1 for RMS error)

(see figure 2 for final product)

Part 2: Image-to-Image Rectification

In this part of the lab, we geometrically corrected a Landsat TM image of the Sierra Leone area to a geometrically correct image of the same area. By distributing GCPs evenly across the image, the software can stitch the new image to the correct image's features.  (see figure 3 for RMS error)(see figure 4 for final product)

Results

Figure 1. For the first geometrically corrected image, I got the RMS error down to 0.3280. We were aiming to get an error of 0.5 or less, so this represents a success.

Figure 2. Overlaid satellite image over map with correct geometric alteration.

Figure 3. RMS error of 0.1757. I consider this transformation a success.

Figure 4. Geometrically corrected image overlaying the original reference image. As you can see, it is a bit hazy, but it is geometrically matched with the first image.

Data Sources

Satellite images are from Earth Resources Observation and Science Center, United States Geological Survey.

Digital raster graphic (DRG) is from Illinois Geospatial Data Clearing House.