GEOG 751: Digital Techniques of Remote Sensing

Exercise #8

Spectral Enhancement: Band Ratioing and Image Filtering

Objectives

To introduce band ratioing techniques.

To learn and understand image filtering techniques.

Part I - Band Ratioing

CASI-2 Image

File - swift_casi_2000-07-01_subset.img

Location - Swift Creek, North Carolina

Date - July 1, 2000

Snapshot - RGB = 10,7,3 over Aeroscan LiDAR canopy.

Compact Airborne Spectrographic Imager-2 (CASI-2)

Band 01 = (484.0nm +/- 25.4nm) - Blue
Band 02 = (529.9nm +/- 20.8nm) - Green
Band 03 = (568.6nm +/- 18.1nm) - Green
Band 04 = (606.5nm +/- 16.3nm) - Green
Band 05 = (637.8nm +/- 15.4nm) - Red
Band 06 = (671.2nm +/- 12.6nm) - Red
Band 07 = (698.0nm +/- 12.6nm) - Red
Band 08 = (721.0nm +/- 08.8nm) - Red
Band 09 = (737.3nm +/- 07.8nm) - NIR
Band 10 = (784.4nm +/- 14.6nm) - NIR
Band 11 = (860.7nm +/- 15.6nm) - NIR

Band ratioing is a process by which brightness values of pixels in one band are divided by the brightness values of their corresponding pixels in a another band in order to create a new output image. These ratios may enhance or subdue certain attributes found in the image, depending on the spectral characteristics in each of the two bands chosen. Begin by displaying the swift_casi_2000-07-01_subset.img image using RGB = 9, 5, 2 and with the No Stretch option checked; then find and select the Image Interpreter button on the Imagine icon panel.

In the menu list that appears find and select the Utilities... option. In the Utilities option window select Operators... In the two empty input windows select swift_casi_2000-07-01_subset.img as the image and in the empty output window add a filename of your choice. Under the input files, select the bands you wish to use. For instance, if you wanted to do a 9/5 band ratio you would select layer nine for input file #1 and layer five for input file #2. Select the Operator to be used, in this case the division symbol. Leave all other fields in their default values. Select OK in the Two Input Operators window. Do this for each of the ratios listed below, being sure to give them an easily distinguishable name.

1) Band 9 / Band 5
2) Band 10 / Band 1
3) Band 11 / Band 2

To combine these into one file so that they can be viewed as a three-layer image, it is necessary to next select the Layer Stack option in Image Interpreter's Utilities menu. The Utilities menu should already be open from the previous work. Place the first ratio image (9/5) that you created in the Input File space by clicking on the open file button and give the Output File a name such as swift-ratio-layerstack.img. Now click on the Add button and you should see the path and name of the first ratio image in the window. DON'T click OK until you have entered all three images. Continue going to the Input File again but this time add the name of the second ratio image (10/1). Select Add and complete the process by adding the last ratio image (11/2). If you mess up, just click the clear button and start over. Leave all other fields in their default settings. Once you see all three files in the window you may now click OK.

You can now display the new ratio layer stack image as a color composite as well as viewing the three individual ratio layers. Do this by opening a three layer arrangement under Open - Three Layer Arrangement under the File menu in the viewer. In the blank window select the ratio layer stack output file you created and choose RGB = 1,2,3 in True Color, and select the OK button. In the three smaller Imagine viewers that display the gray scale results of the band ratios, look in the title bar of the viewer to determine the layer. If you ordered them correctly in the Layer Stack section the layers should correspond to the ratios listed above, i.e. layer 1 should be the 9/5 ratio. The composite will probably look unfamiliar to you. A roving box should also appear in your color composite window. This window can be resized with the mouse and corresponds with the area displayed in the three small viewers. The box can be dragged around the larger viewer as an additional query method.

1) Turn in a brief statement of what information specifically is highlighted in each of the band ratios.

9 / 5 Ratio:
10 / 1 Ratio:
11 / 2 Ratio:

2) Using band ratioing techniques, does a high or low correlation between bands extract the most information? Why?
3) Based on the band ratios you have just performed, which one most enhances vegetation and vegetation differences?

Part II Image Filtering

Landsat TM Image

Quickview	File - santee.img
	Band 1 = Blue (.45-.52) Band 2 = Green (.52-.60) Band 3 = Red (.63-.69) Band 4 = NIR (.76-.90) Band 5 = MIR (1.55-1.75) Band 6 = MIR (2.08-2.35

This section of the exercise deals with image filtering and uses the Landsat TM imag called santee.img. You will select a subset (portion) of santee.img and apply a series of filters to it. To best appreciate the effects of these techniques you should choose an area that has a variety of lines or edges present (i.e. the edge of a forest, road, wetland boundary, land/water boundary etc.). You might want to create a separate folder to house all of the files you create in this part of the exercise.

Subset the image the same way we have done in past exercises. If you have forgotten how to do this, here are the steps. Make sure the entire santee.img is displayed in a viewer and select the right mouse button (rmb) inside the viewer to bring up the Quick View utility menu. In the Quick View menu select Inquire Box. In the menu that appears change the map coordinates to file coordinates. The white box that Imagine places over the image shows the extent of the area that will be subseted. You can move the box with out changing its dimensions by placing the cursor inside the white box and, while holding down the lmb, moving the box to another location. You can change the dimensions of the box by holding down the lmb while the cursor arrow is on an edge or side of the box and, then draging the cursor around (the box should follow). You can also change the dimensions of the box by directly entering the row and column values in the menu that appeared with the white box.

Once you have positioned your box in the area to be subseted, click on the DataPrep button in the Imagine icon panel. In its menu find and select the Subset option. Use the following directions to fill in the appropriate menu choices and note that we will make an individual subset for each band.

Select santee.img as the input file.
Name the output file (it might be helpful in keeping track with each of the many subsets you will do by using output filenames that are descriptive, i.e. santeeband1, sanband2, sband6, etc.).
Select Coordinate Type equal to File then click on the button that says "From Inquire Box". Notice that your inquire box coordinates have been automatically included in the spaces that determine the boundary of the subset.
In the lower part of the window find the words Select Layers. In the space to the right you will find the entry "1:6", this means layers 1 through 6 will be included in the subset (the default is to include all image layers). You want to extract each individual layer as a separate file, so change this entry to read only the desired layer (i.e. if you want to extract only layer 2, type 2 in the entry).
Select OK and the subset process will begin.

Do this for each of the six bands (i.e. santee1.img, santee2.img, santee3.img, etc.).

Now you will select the filter type to use on the subsets. Under the Image Interpreter menu, select the Spatial Enhancement option and then the Convolution option. This opens a window which allows you to select from a variety of existing convolution matrices or create your own. The size of the matrix (3x3, 5x5, 7x7 etc.) is also chosen here. To filter an image subset do the following:

Select the input file for convolution (start with santee1.img).
Enter an output file name for the resulting filtered image (it might help you to include in the filename the type of filter that was used i.e. - santee1hpf7x7.img meaning santee. img, band1 subset, hpf for high pass filter, and 7x7 for the matrix size).
Under Kernel you have some default filter types and sizes available for use. Most of the kernels you will need can be found under this menu. To create your own filter, (you will need to do this for both the Laplacian and the compass gradiant filter) select New below the Kernel window. An empty Kernel will open that has the title "(untitled)". Under File, in the Kernel window, select Librarian. The Kernel Librarian window should appear. Scroll to your directory if it is not already there (it probably is not) and type in the Library filename space a name for your own kernel library. An example would be my-kernel.klb. When you have created your own kernel library make sure you give it a name and a description then click Save. Make it the active kernel library by selecting it. The kernel should automatically receive the name you specify into the Kernel Editor. You can modify the kernel simply by typing in the cells in the Kernel Editor . When you are done with your changes select Save in the Kernel Librarian window and Close the window.
Select the Fill option for Handle Edges By.
Select the OK button and the filtering process will begin.

Below are six types of filters that need to be run on each of the six bands. Choose a feature of interest in the image and see how it changes with the passing of each filter. Use your knowledge of spectral reflectance characteristics to answer the questions below.

a. A 3x3 Low-Pass filter
b. A 7x7 Low-Pass filter

c. A 3x3 Edge Enhancement filter

d. A Laplacian filtered image using the following matrix values (Notice the similarity to a high-pass filter):

1 -2 1

-2 4 -2

1 -2 1

e. Any size High-Pass filter

f. Design a Directional Compass Gradient filter to enhance lines running SW-NE (see the textbook for help).

To answer the following questions it might be helpful to have each of the filtered images and the composite image in viewers for quick reference.

1) Which of the following would make an image blurrier, a 3x3 or a 7x7 Low-Pass filter? Why?
2) What edges are highlighted with the 3x3 Edge Enhancer? Is anything else enhanced as well?
3) What does the Laplacian filter tend to enhance and/or supress in the scene?

4) What is the result of performing a High-Pass filter on an image?

To complete the rest of the exercise, create and print a new map composition containing four images: choose any of the three filtered images you created as well as an example of the image you filtered with your uniquely designed filter. Be sure your map composition has appropriate annotation. The items identified in the previous labs should be used as a guide for how this composition will be graded.

Back to Exercises

Last Modified: October 18, 2004