Lab 6: Geometric Correction

Goals and Background:

The goal of the lab is to develop knowledge and first-hand skills in geometric correction. Geometric correction is an image preprocessing exercise that is used on satellite images prior to any data extraction from the satellite images. Skills learned in the lab will be useful to future work with remotely sensed images.

Methods:

Part 1:

Geometric correction of the Chicago_2000.img Landsat TM image (United States Geological Survey, 2015) was performed using a United States Geological Survey 7.5 minute digital raster graphic of the same area (Illinois State Geological Survey ,2015). This was accomplished by adding ground control points (GCP's) on both the input and reference images using the Multipoint Geometric Correction dialog box in ERDAS IMAGINE 2013. A first order polynomial equation was used to rectify the image. Four GCP's were added to each image and a total root mean square (RMS) error of 2 or lower was achieved by replacing the GCP's in more accurate locations. The Multipoint Geometric Correction dialog box with an appropriate RMS error can be seen in Figure 1. The geometric correction was performed after the RMS error was lowered to 2 or less. The Display Resample Image Dialog button on the Multipoint Geometric Correction tool bar was used to complete the geometric correction process. The Chicago image was geometrically corrected using nearest neighbor interpolation and accepting the other default values. The results of the geometric correction for the Chicago image can be seen in Figure 2.

Part 2:

Geometric correction of the geometrically distorted Sierra_Leone_east1991.img   was performed using a corrected Landsat TM reference image of the same area called Sierra_Leone_east1991grf.img (United States Geological Survey, 2015). The correction process was identical to the process carried out in Part 1 except for a couple exceptions. First, a third polynomial equation was used to correct the image.  Second, twelve GCP's were used. Third, an RMS error of 1 or lower was achieved to correct the image (Figure 3). Finally, bilinear interpolation was used when completing the geometric correction process. The results of the geometric correction for the Sierra Leone image can be seen in Figure 4.

Results:

Figure 1: Total RMS error for Chicago_2000.img is less than 2.0.

Figure 2: Geometrically corrected Chicago image using nearest neighbor interpolation.

Figure 3: Total RMS error for Sierra_Leone_east1991.img is less than 1.0.

Figure 4: The geometrically corrected image for Sierra Leone, which used bilinear interpolation, is relatively spatially accurate.

References:

Illinois State Geological Survey. (2015). [Digital raster graphic is in img. format]. Illinois Geospatial Data Clearing House. Retrieved from http://www.isgs.uiuc.edu/nsdihome/

United States Geological Survey. (2015). [Satellite images are in img. format]. Earth Resources Observation and Science Center. Retrieved from http://eros.usgs.gov/