Geometric Correction

Lab 6

 

 

 

Introduction

 

This week in lab we focused on the idea of Geometric Correction. Geometric correction is the removal of geometric distortion in an imagery so that individual pixels can be in their proper planimetric (x,y) position. This lab was structured to develop your skills on the two major types of geometric correction that are normally performed on satellite images as part of the preprocessing activities prior to the extraction of biophysical and sociocultural information from satellite images.

 

 

 

Methods

 

 

Part 1

 

First open up ERDAS Imagine with a viewer. In this viewer bring in the image Chicago_drg.img. Then open a second viewer and bring in the image Chicago_2000.img. Click Multispectral and then on Control Points. This process can be seen below in Figure 1.

 

 

Figure 1: This is a screenshot of the process of selecting the Control Points tab.
Image created by: Cyril Wilson

 

 

 

Now the Set Geometric Model window will be open, under Select Geometric Model select Polynomial and then hit Ok. Another window will pop up, accept the default. Then navigate to the Lab 6 folder and add your reference DRG image - Chicago_drg.img. Then click Ok on the Reference Map Information.

 

The Multipoint Geometric Correction window will open up two panes. On the left is the input image (Chicago_2000.img), while the reference image is on the right pane (Chicago_drg.img). Each of these panes contains three windows. The top left and top right panes show the entire input and reference images respectively. The other two central top panes shows the areas that are zoomed into on the input image and also that zoomed into the reference image.

 

Now starts the process of adding GCPs to the images. But first the default GCPs need to be deleted. Once this is done the task is to add four pairs of GCPs on the images. Even though only three are needed, extra can't hurt and will give a better fit.

 

Begin by clicking on the Create GCP tool at the top of the page. Then add the four GCP points in the locations exactly like shown in Figure 2.

 

 

Figure 2: This is a screenshot of the instructed areas as to where we were supposed to place our sets of GCP points.
Image created by: Cyril Wilson

 

 

Now is the process of making sure you're doing a good job by evaluating your GCPs. This is done by looking at the Root Mean Square (RMS) error. This is shown at the bottom of the page. For this lab the goal is to reach a RMS of below 2.0. This is done by repositioning the GCPs in their correct placed on both images. After all of the RMS values are below 2.0 now comes the process of creating the corrected image begins.

 

 

Begin this process by clicking on the Display Resample Image Dialog. Name the output file Chicago_2000gcr.img and place it into your Lab 6 folder. click ok and then bring the resampled image into your ERDAS viewer.

 

 

 

Part 2

 

This section is essential going to be the same as the first part, with a few changes. The image that will be located in the first viewer will be sierra_leone_east1991.img and Sierra_Leone_east1991grf.img in the second viewer. Click on Multispectral and Control Points again. Click on Polynomial under Select Geometric Model, then click Ok. Set your reference image as Sierra_Leone_east1991grf.img, and click Ok. On Polynomial Model Properties change the polynomial order to 3 and click close.

 

Now begins the process of adding the GCPs again, but this time 12 are going to be added, even though only ten are needed. The GCPs will be located in the exact same location as of those seem in Figure 3.

 

 

 

Figure 3: This is a screenshot of where the GCPs should be located for this section of the lab.
Image created by: Cyril Wilson

 

 

Next click on the Display Resample Image Dialog on the Multipoint Geometric Correction toolbar. Name the output file sl_east_gcc.img. Change the resampling method to bilinear interpolation and click Ok. Now bring the resampled image into the ERDAS viewer.

 

 

 

Results

 

For the results for each section I either can to upload an image or answer a question or two. I will show you my created images and answered questions for each section below.

 

 

Part 1

 

 

 

 

 

 

 

Q1: The Chicago_drg.img image was the image used in an image-to-map rectification. The image data pixel coordinates were rectified/transformed using the map coordinate counterparts. The rectification process involves converting a data file coordinate to some other grid and coordinate system known as a reference system. So this particular image was used as a reference to transform the map.

Q2: The resampling method used above is called spatial interpolation. Spatial interpolation is the use of GCP pairs to establish a geometric coordinate transformation that is applied to rectify the location of pixels in the output image with a value from a pixel in the unrectified input image.

Q3: GCPs must be dispersed throughout the image to achieve a reliable geometric correction.

Q4: A first-order polynomial equation/model is used to fit a plane to the data. This is the fitting of polynomial equations to the GCP data using least-squares criteria to model the corrections directly in the same image domain.

Q5: The minimum number of GCPs needed for a first order polynomial would be 3.

 

 

Part 2

 

 

Q6: UTM (Zone 29)

Q7: 10 GCPs

Q8: Because in part 1 only 3 GCPs were necessary, in this section 10 GCPs are necessary. Currently there are only 9 GCPs, so it will display Model has no solution

 

 

 

 

 

 

 

 

 

Q9: The new image has a greater geometrical correctness. Even though there is still cloud cover in the image it is still easier to identify features on the new image. The old image was much more distorted and had a lot more cloud cover.

Q10: Bilinear interpolation is more spatially accurate than nearest neighbor. This image was in more need of greater spatial accuracy than the image in part one, that is why bilinear was used instead of nearest neighbor.