Lab 6: Geometric Correction


Goals:

The goal for this lab was to learn how to use ground control points (GCPs) and a reference image to make geometric corrections to satellite images.

Methods:

Part 1:

The first part of this lab used a distorted satellite image of the city of Chicago with a digital raster graphic (drg) of the city of Chicago that is spatially correct. The drg is the reference layer that the distorted image will be corrected to match. First, the distorted satellite image needs to be brought into the work space and under the multi spectral tab, use the control points tool. This will open the geometric corrections tool. When the tool opens, set the geometric model to polynomial. Next, the reference layer needs to be added to the tool. For this part of the lab the drg is the reference layer. After the reference layer is in the tool, the order of polynomial desired is selected. For this part, a first order polynomial is used. Next, two viewers will be viable each containing either the distorted image or the reference layer. Once this appears, GCPs can be placed in the same location in each image. For a first order polynomial, only 3 GCPs are required. Once the GCPs are placed, the tool will provide the RMS value for the image. This tells the amount of pixels the error is allowed. For this part a RMS of 2 or lower. If the RMS is above 2, each GCP can be adjusted individually to decrease the RMS. When the RMS is under 2, the tool can be run using nearest neighbors re-sampling method to create a geometrically corrected image.

Part 2:

The second part of this lab follows the same steps as part one, but with some small changes. The distorted image is of eastern Sierra Leone in 1991 and the reference image is a geometrically correct image of Sierra Leone. Follow the same steps as part one until the polynomial order is selected. For this part, a third order polynomial is used which requires at least 10 GCPs. This part uses 12 GSPs just to be sure. Once all the GCPs are placed and the RMS is under the acceptable threshold, in this case it will be 1, and the interpolation method is set to bilinear interpolation instead of nearest neighbor. Finally, the tool can be run to create a geometrically correct image of the area.

Results:

Part 1:

The results for part of the lab resulted in a geometrically corrected satellite image of Chicago (Figure 1) with a RMS value of  0.2063 (Figure 2). This means the image has a margin of error of 0.2063 pixels.

Figure 1. Corrected image of the City of Chicago. 


Figure 2. RMS value for individual points and a total RMS value.

Part 2:

The results for part two of this lab is a geometrically corrected image of eastern Sierra Leone (figure 3) with a RMS value of 0.3874 (figure 4). This means the corrected image has a margin of error of 0.3874 pixels.

Figure 3. Corrected satellite image of eastern Sierra Leone. 
Figure 4. RMS values for individual points and a total RMS value.


Source:

Earth Resources Observation and Science Center, United States Geological Survey.

Illinois Geospatial Data Clearing House

Comments

Post a Comment

Popular posts from this blog

Lab 8: Spectral Signature Analysis and Resource Monitoring

Image
Goals: The goal for this lab is to become familiar with measuring, collecting, and interpreting spectral reluctance and signatures. The other goal for this lab is to learn how to monitor resources with NDVI and ferrous mineral monitoring. Methods: Part 1: The first part of this lab deals with spectral signatures. The first thing that is done, after a photo is loaded in to program, is to collect the spectral signatures for various land covers. This is done by identifying the land cover type desired in the image and creating a polygon around it. Next, select the supervised drop down and click the signature editor. Once this is open, the spectral signature for the area inside the polygon can be found. These step can then be repeated until all the desired land cover spectral signatures are found. Part 2: Section 1: The first section of the second part of this lab deals with calculating vegetation density using satellite images. This is done by using the desire satellite image...
Read more

Lab 5: LiDAR remote sensing

Image
Goal: The Goal for this lab was to become familiar with LiDAR data and to use it to make various products such as DTM and hillshade models. Methods: Part 1: For the first part of this lab, LAS data was imported into Erdas Imagine so the LiDAR point cloud can be visualized. This is done by selecting the LAS files that are desired and to open them in Erdas Imagine. This will bring in the point cloud into the workspace and allow it to be visualized. Part 2: For this part of the lab, a LAS dataset needs to be created and to make sure that the quality of the data is good by looking at the statistics. To do this, arcmap needs to be opened and the folder the LAS data is in needs to be connected to arcmap. Next, in the catalog, a new LAS dataset needs to be created by right clicking on the folder its going into and selecting new and then LAS dataset. Once the LAS dataset is created and named, the individual LAS files can be brought into the newly created dataset. Once the files a...
Read more