ERDAS Imagine Image Processing Lab


There is a Landsat 7 image of Mount Desert Island (MDI), Maine located in /scratch/maine02/mdi.051900/mdi.l7.051900. The image name is mdi.L7.051900.img. Make a directory in your home directory on everest (U:) called maine02. Then make a subdirectory called images. Copy the MDI image to the U:maine02/images directory.

Start ERDAS Imagine version 8.5: Start --- Programs --- ERDAS IMAGINE 8.5 --- ERDAS IMAGINE 8.5

An icon panel will appear with 13 buttons, Viewer, Import, DataPrep, etc.

A Viewer window will also appear. This is where the software will display a digital image.

In the Viewer window there is a Menu Bar with 9 dialog box headings; file, utility, view, etc. The Tool Bar is just below the Menu Bar and contains many icons.

Left-click on file and an icon menu appears. Left-click on open and then left-click on raster. A raster icon menu will appear. You will be displaying a raster Landsat 7 image of Mount Desert Island. Navigate to /scratch/maine02/mdi.051900/mdi.l7.051900 to read in the image.

To look at scene statistics select the Show Information icon on the ToolBar.

imagine.8.5.viewer.jpg (40437 bytes)

An ImageInfo window will appear for Layer 1. You can change the ImageInfo to each layer or band of the displayed image.

Note that in the Layer info: portion of the dialog box it tells you there are 6 image layers. These layers correspond to the 6 bands of the electromagnetic spectrum that Landsat 7 images, three bands in the visible, three bands in the near IR and one band in the thermal IR at 30m ground resolution. There is a 7th panchromatic band at 15 m ground resolution that I did not include in the image file.

To display a color digital image you need three, and only three, bands. We can select any three of the seven. By default band 4 (IR) will be assigned the color red, band 3 (visible) will be assigned the color green and band 2 (visible) will be assigned the color blue. If a near IR band is assigned the color red, then anything that is moderately to highly reflective in the near IR band will show up red. In fact, vegetation, in general, is moderately to highly reflective in the near IR and so will display as red.

Read in the manhattan image again (File --> Open --> Raster Layer) but when the dialog box appears, select Raster Options and change the bands assigned to RGB. Note the color changes in the manhattan image.

Also note in the Layer info: portion of the dialog box that the image is 713 rows of pixels by 848 columns of pixels by 7 layers. Including just data, how big should the file be when stored on a computer?

Now take a look at the histograms for bands 4, 3, and 2. Open three ImageInfo dialog boxes and click on the histogram icon for each. Display the histogram for layers 4, 3 and 2. Compare and contrast the histograms for bands (layers) 4, 3 and 2.

On separate paper type your answers to these questions. Draw diagrams if appropriate.


Looking at bands 2, 3 and 4 answer the following questions;

1- What is the minimum and maximum brightness value (BV) for each band?

2- What is the mean value for each band.

3- Is the histogram a bell-curve or multi-modal?

4- What is the size of each band in bytes? How much disk space would a seven-band image of this size need? Show your calculations.

Now display each individual band (2, 3 and 4) in separate Viewers. The individual bands will be monochrome. To display band 2 in the Viewer dialog box, select the number 2 for the Layers To Color. That is, put the same band through Red, Green and Blue filters. Arrange your four images, the color composite, and the three monochrome of bands 2, 3 and 4, so you can see all for at the same time and answer the following questions;

5- Look at the three b&w images. Which band is best at discriminating between land and water? Why? What part of the EMS does this band represent?

6- Which band best shows vegetation? What part of the EMS does that band represent?

7- Of the three b&w images which shows the most atmosphere haze? Why?