Lab One, For 753, NC State

In this lab we will introduce the data to be used throughout the course. We will also start using the IMAGINE image processing software package to learn some basic operations.

There are three main sources of image data we will use in the image processing labs.

We will use several TM images in the lab. Most of these images will be subset images cut from an entire Landsat "scene". The subset images are roughly 1000 pixels by 1000 pixels, while an entire Landsat scene is about 6500 by 6500 pixels. Likewise our subset images are about 6.5 megabytes while a full scene is about 250 megabytes. TM data are 28.5 meter spatial resolution with seven multispectral bands.

We will also use the French SPOT satellite data. SPOT panchromatic images have a 10 meter spatial resolution and one band. SPOT multispectral images have a spatial resolution of 20 meters and five multispectral bands.

Digital Orthophotographs were acquired from the United State Geological Survey's Digital Ortho Quarter Quadrangle (DOQQ) program. These data have a nominal spatial resolution of one meter and were developed by digitizing National Aerial Photography Program (NAPP) photos. DOQQs are arranged by USGS 1:24,000, 7.5 minute, quadrangle and there are four DOQQs per 7.5 minute quad.

In addition to image data we will also use some vector data. All of these are Arc/INFO coverages. We will also use some Arc/GRID format data. As you will see, these are not vector data types (they are raster), but we include them with the vector data since they were produced with Arc/INFO.

All of the image and vector data are projected into the Alber's Equal Area Conic projection using the following U.S. Federal standard parameters:

Projection: Albers
Units: meters
Datum: NAD83
Spheroid: GRS80
1st Standard Parallel: 29.5 deg.
2nd Standard Parallel: 45.5 deg.
Central Meridian: -96.0 deg.
Projector's Latitude: 23.0 deg.
False Northing: 0 deg.
False Easting: 0 deg.

Much of the class data were acquired through various project done here at CEO, including the National Oceanic and Atmospheric Administration's (NOAA) Coastal Change Analysis Program (CCAP) and the accuracy assessment of Regions IV and V of the Multi-Resolution Land Characteristics (MRLC) Consortium's National Land Cover Data (NLCD).

We will be using ERDAS' IMAGINE image processing software running on Windows NT workstations. Image data tend to use a lot of space. As we go through the labs you will create different manipulations of the original images. Please delete all "test" or temporary images you create. Also, please do not change Preferences or Configuration parameters. Before you end your session, confirm that you really want to quit IMAGINE and answer "no" to the question whether you want to print the LOG file, if the question appears.

In this and other labs, Windows commands or IMAGINE application names which you must type or select are shown in bold face. Variable or filenames are shown in italic. Name1/name2/name3 means a continuous operation, selecting an option (name1) first from a "parent" menu and then from subsequent "child" menu(s) (/name2/name3). Note: There is on-line help in ERDAS IMAGINE. Just use the Help menu (or press F1) wherever you are to get the help page about the current menu or dialog box.

You start Imagine by selecting Start/Programs/Leica Geosystems/ERDAS Imagine 9.0/ERDAS Imagine 9.0. If you are still having trouble ask the lab instructor for help. Once you get into Imagine it will give you its "main menu" bar and one "viewer" by default. These are shown below:

One Viewer is automatically created when IMAGINE loads. To start an additional IMAGINE Viewer, left-click the Viewer button in the IMAGINE main toolbar (once for each viewer). A Viewer menu bar appears across the top of every Viewer window. You can left-hold on each word on the menu bar to access a pull-down menu of IMAGINE operations that are performed in a Viewer. Below the menu bar is the tool bar, which includes icons for fast access to frequently used functions). Select File/Open/Raster from the menu bar, or press the first button on the tool bar, to get the Open Raster Layer dialog box (you may have a different list of files than what is shown).

Use this dialog box to select the image and type of data display for the viewer. In the Files of Type box, IMAGINE Image is the system default file type. All the files in your directory with the extension specified under File Type will appear in the list of files. You can left-hold the Files ofType popup to select a new default file type. (We will be working mainly with IMG files.)

Note: All files used in the labs will be found on the server: gettysburg.nrrc.ncsu.edu in the "for753" folder. Map this location as a network drive (\\gettysburg.nrrc.ncsu.edu\for753) using the username and password "for753".

Use the file list to select the input file. We will first work with the file raleigh94_alb.img. This file is in the tm_data folder. Left-click once on the file name to select that file. Now click the Raster Options tab. You will see a dialog box like this:

Notice that certain default parameters are automatically set when you select a file. If the image to be displayed is a multilayer file (spectral data with more than one band), you can choose any three bands to display in any order. Usually, IMAGINE will recognize the scanner type (TM, SPOT, MSS, etc.) and default to display a standard composite. You can change the bands displayed by using the up/down arrows next to the "Layers to Colors: Red Green and Blue channel displays.

Other default parameters can also be changed before displaying the image. For example, change the following display options to get an image or a desired portion of an image displayed at a different zoom ratio.

Set View Extent allows you to specify the upper left and lower right coordinates of the portion of the image you want to display. When Set View Extent and Fit to Frame are off, you can enter a zoom ratio (enlargement or reduction) for the data (the default is 1, meaning every pixel of the image is displayed in one picture element of the Viewer). When the image is reduced, the magnification factor is less than 1.000, and when the image is enlarged, the magnification is greater than 1.000.

Left-click on the OK button to display the image (which will automatically close the Open Raster Layer dialog box).

In the IMAGINE Viewer, select View/Create magnifier, or use "Quick View"/Zoom/Create Magnifier by holding the right button of the mouse anywhere inside the Viewer. A small window will appear and, on the image, a small outlined box will indicate the area shown in the magnifier. Pixels inside this box are magnified (default 2X) in the small window. Place the cursor on the middle point and hold the left button of the mouse to move the magnifier, or place the cursor on the edge points to stretch the magnified area to any desired size (note this changes the zoom ratio). You can also access the Quick View menu by right-holding the mouse in the magnifier window.

The Inquire cursor dialog box gives information about individual pixels by using a cursor which is displayed as a cross-hair in the Viewer window. An image must be displayed in the Viewer before you inquire cursor . Left-click View/Inquire Cursor from the Viewer menu bar. The Viewer window which is under inquiry will be listed in the Inquire Cursor dialog title bar. To move a cursor (cross hairs), you can either left-hold the intersection of the cross-hairs and drag it to the desired location, or enter the desired coordinates in the Inquire Cursor dialog box, or use the 4-way "nudgers" (arrows) in the Inquire Cursor dialog box to move the cursor in any direction. The inquire cursor dialog box shows both the coordinates related to where the cursor is positioned and the digital numbers for the image data at that location. The inquire cursor is good for finding out detailed image information for a particular point.

By "Global" information, we mean information related to the entire image as opposed to any particular location. Select Utility/Layer Info... from the viewer menu to get the Image Info dialog box:

The set of tools in "image info" allows you to view, edit, and print most of the information of an image file (.img file). Caution: the information in this dialog should not be modified in this class, except as directed in subsequent exercises, because IMAGINE uses this information as input data.

Information is displayed for a default data layer (layer #1, i.e. band #1). You can change the layer number to get the information for a different data layer if your data is multi-band data. Note that layer numbering is sequential in an IMG file and may not coincide with band numbers for a sensor (e.g. an IMG file may contain TM bands 2, 3, and 4 which, in the file, would be layers 1, 2, and 3, respectively).

In the image information dialog box you can see the statistics related to the band being displays. These statistics pertain to the entire image. For example, the "mean" refers to the mean of all the pixel values for the given band.

Use View/Histogram from the Image Information menu bar to view the histogram for the current data layer whose information is displayed in the Image Info dialog box. Explore other band information about the raleigh94_alb.img image file.

The histogram gives additional detail beyond the global statistics. It show the spread and shape of the how the pixel values are distributed over all of the possible values for that particular band. The x-axis represents the digital number and the y-axis shows how many pixels in the image have that particular digital number. If you are having trouble understanding the information contained in the histograms, please ask the lab instructor to explain.

Some things to try...

1. Use Utility/Image Info utility to obtain mean and standard deviation for bands 1 - 6 from the tm_data/raleigh88_alb.img file.

2. Using File/Open/Raster in the viewer menu bar, display at least six different false color composites. Try some of these common composites.

a. Standard false color composite (4,3,2)
b. True color composite (3,2,1)
c. False color composite (4,5,3)

Be prepared to briefly discuss at least one of the composites which you display. Factors to consider include: what bands were used (and in which display channel), the appearance of specific ground features in each composite (e.g. what color is vegetation?), or whether the composite appears to enhance any particular feature.

3. With the tm_data/raleigh94_alb.img image open in a viewer, select File/Open/Vector layer . This will allow you to open a vector file. Open the Arc coverage called streets_a_c. Click on this file and click Ok (make sure the "Clear Display" is NOT ON. When the coverage opens, select Vector/Viewing Properties. This will give you the following Dialog Box:

Left-click on the small button to the right of "Arcs". This will give you an options box. choose "Other." Then, in the next dialog box that comes up, right-click on the Outer Color: box and select the color yellow. Then click Ok This will close the options box. Now left-click on Apply. You should get yellow street lines isplayed over your image. Close the "Properties" box (and answer No to the question about saving the changes to a symbology file).

Try to find your house
Try to find NC State
Locate the beltline
Locate Capital Blvd.

Right-hold on the Vector: streets_a_c coverage and select Delete Layer and then left-click on Apply. This should bring the viewer back to just the raleigh94_alb.img image.

5. Now with the raleigh94_alb image open in your viewer, select File/Open/Raster... and select the tm_data/raleigh88_alb.img file. Choose the "Layers to Colors" so to match the image currently displayed. Then left-click on the "Clear Display" box so that it is OFF. Now left-click on OK Both images are now currently in your viewer and your "Arrange Viewer" box should show the two image files.

Select Utility/Flicker... Repeatedly left-click on the Manual Flicker button.
Left click once on the Auto Mode box. Try Changing the speed.
Close this box and experiment with Utility/Blend and Utility/Swipe

6. Using the "Arrange Layers" dialog box, clear the raleigh88_alb layer then open the file spot_data/raleigh_pan96_alb.img Now try the Blend, Swipe, and Flicker Utilities.

Using the "Arrange Layers" dialog box, clear the spot data layer then open the file doqq/rale_se.img Again try the Blend, Swipe, and Flicker Utilities.

What is different about opening the SPOT and DOQQ image? (Hint, the 94 TM data has seven bands while the SPOT and DOQQs have only one band).

1. Why do trees appear red in "standard false color composite" Thematic Mapper images?

2. What band combination in a Thematic Mapper image could you use to make trees appear blue?

3. The Histogram tab in Utilities/Layer Info shows a histogram (number of pixels per bin) of the digital numbers of the individual bands in an image. Causing certain ranges of digital numbers to be hidden when an image is displayed is called thresholding. Conceptually, how would you threshold the TM image raleigh94_alb so that no clear, deep water is displayed. In other words, which digital number value(s) (high, medium, or low) of which band(s) would you hide so that no clear, deep water shows up.

4. Select a Thematic Mapper image of your choice to open. In the "Raster Options" tab of the Open dialog box, change the "Dispay as:" dropdown box to "Gray Scale". Select Band 4 (a near infra-red band). Then check the "Data Scaling" box. Click OK. In the next dialog box, move the two small arrows to arbitrary locations along the x-axis of the histogram. Proceed with opening the image. Repeat this procedure using different arrow settings. NOTE: If prompted to save the image, choose "No." What is this "Data Scaling" option? What are you doing when you move the arrows?

5. Select a Thematic Mapper image of your choice. In the "Raster Options" menu of the Open dialog box, check "No Stretch". Why does the image look different? (Hint: Use the View / Inquire Cursor tool and examine the difference between the "File Pixel" vs. "LUT Value" with and without the "stretch". Note: LUT stands for "Look Up Table".)

6. Of the types of image data that we have talked about so far, which would be best for a forest type classification of a 1000 acre plot? Why?

7. Which data type would be best to use as a basemap to show the locations of traffic lights in Raleigh? Why?

8. Use the Imagine Help system to briefly explain what a "pyramid layer" does.

9. Why is the sky blue?