Remote Sensing Lab 6: Geometric Correction

Goal

The goal of this lab was to introduce us to geometric correcting of a satellite image. This lab is focused on the two main ways to correct a satellite image, which will be introduced to in the method section of this blog post.

Method

In the first part of the lab we are working with two Chicago images. One is a topographic map of the Chicago and surrounding area, while the other one is a remotely sensed image of a smaller portion area. The two images should be in separate viewers on Erdas Imagine 2013. The next step is to select the multispectral tab and click the tool Control Points. In this lab we will use the Polynomial function and leave all the other options at their defaults. After following through the next few pop up boxes you will come to a place where we will have to select the reference image, which will be the Chicago_2000.img. Since we are using the 1st polynomial function, we will need at least three points before a solution will be possible. When placing ground control points the fourth one will place automatically. After placing the ground control points, we will have to move them around to minimize the Root Mean Square error (RMS). Ideally you would like the RMS number to be less than 2.0. Once this is done with all four Ground Control Points you will hit the Windows looking logo button to finalize the image.

Geometrically Corrected Image

Part two of this lab is to work with two images of Sierra Leone, instead of working with one map and one image. This process is done similar to the way of the first part, but instead of selecting the 1st polynomial we will be selecting the 3rd polynomial. This way we will now need 10 Ground Control Points before a solution can be found. After moving around the Ground Control Points to make sure the RMS error is below 2, you will hit the Windows looking logo button again to finalize the image. Unlike in the first part when we used Nearest Neighbor as the default, we will be using bilinear interpolation to fix these images.

Corrected Sierra Leone images with RMS errors present

Results

For part one the results were a geometrically corrected satellite image of the Chicago area. It was correct when zoomed in between the two images.
For part two the images were a lot closer than the previous. They were much more geometrically correct than the previous images.

Sources

United States Geological Survey (USGS) 7.5 minute digital raster graphic (DRG)
Images provided by Dr. Wilson