The goal of this lab is to display the technique and skills to geometrically correct a satellite image. Two types of geometric correction will be displayed, image-to-map and image-to-image. Geometric correction is necessary because an image may be distorted or not collected in its proper planimetric position. This results in an image that is skewed or objects appear different than they actually are.
Methods:
In order to geometrically correct an image there must be a reference image that is already accurate. The first step is to bring the correct reference image and the image to be corrected into Erdas. Once both images are in Erdas in seperate viewers the process is ready to begin. To correct the image the user must add control points to both images. This is done by using the control points tool under the multispectral raster processing tools. A number of windows then pop up to set the parameters. the reference image must be selected again and the order of polynomial must be selected. In the image-to-map example, first order polynomial was selected. This means only three ground control points are necessary to correct the image. When the parameters are set, a new window will open and the user is ready to add control points. Control points 1-3 must be manually added to both images in the same exact spot. After 3 are added it will say "model solution is current" and control points will automatically be added on the second image when inserted in the first. After all the points are added, some fine tuning of the points is required to lower the root mean square error (RMS error). The lower the RMS value the more accurate the image will be.
For the image-to-image geometric correction, the same process was used but this time was performed using a third order polynomial. This mean ten control points are necessary to correct the image.
Results:
Image-to-Map Geometric Correction Control Points
When adding control points it is important to zoom in when adding each point to match the exact pixel that the point is being placed in both images. The Image on the left is the image that is being corrected and it was corrected to within .4 of a pixel to the reference map.
Image-to-Image Geometric Correction Control Points
A total of 12 ground control points were placed in the above photo. 10 GCP's are necessary in order to correct a third order polynomial correction, but 12 were added to ensure even more accuracy. The RMS error value in the bottom right of the image is .001 and is extremely accurate but not perfect.
Conclusion:
It is necessary to geometrically correct an image before using it for analysis. Erdas makes this process user friendly and easy for everyone to do in a timely fashion. The resulting image will then be ready to take place in research or further analysis. This is a simple tool that should be used every time a new image is being used in a workplace.
Sources:
Satellite images
Earth Resources Observation and Science Center, United States Geological Survey
Digital raster graphic (DRG)
Illinois Geospatial Data Clearing House
No comments:
Post a Comment