Lab 4: Introduction to IDRISI Image Processing

Due: Week of February 12^th (you have two weeks for this lab).

The goal of this lab is to give you the foundations for working with IDRISI, the image processing program that you will be using in all of your subsequent labs, and for your final project. As such, this is a long lab (6 hours is an estimate). However, remember that with this experience you can put down in your resume that you have some working knowledge of image processing.

Given the length of this lab, you should budget your time accordingly. As a goal, you should finish Part 1, Exercises 1.1-1.3 in the first week. In the second week, complete the rest of Part 1, Exercises 1.5 and 1.6, and Part 2. (Note that Exercises 1.4 and 1.7 are optional.)

Marks: 50

Background to IDRISI

In the last lab, and in your image presentations, you have been / will be viewing remotely sensed images that most often have been manipulated by an image processing system (e.g., the production of a 'composite' or false colour image). During this lab, and for the remaining labs (and in your project), you will be working with the image processing software program called IDRISI. IDRISI was developed by Ron Eastman, who is a professor at Clark University (Worcester, MA). IDRISI has been around since 1987; we have used IDRISI in the Department since the early 1990's.

For the remaining labs, you will be using three components of the IDRISI Selva package:

1. The IDRISI software (launched from Start / All Programs / IDRISI Selva; click on IDRISI Selva).
2. The IDRISI tutorial and the associated data sets. The tutorial (PDF) can be found here:
   C:\Program Files (x86)\IDRISI Selva\Documentation\IDRISI Tutorial.pdf.
3. The IDRISI manual (C:\Program Files (x86)\IDRISI Selva\Documentation\IDRISI Manual.pdf).

In completing the next four labs (Lab 4: Introduction to IDRISI Image Processing, Lab 5: Introduction to Image Enhancement and Unsupervised Classification, Lab 6: Introduction to Image Analysis: Supervised Image Classification and Lab 7: IDRISI's Land Change Modeler) you will:

In Part 1: Go through the IDRISI tutorials relevant to the lab topic and answer some questions pertaining to the tutorials, and

In Part 2: Apply the tools and concepts you learned in the tutorials to a dataset specific to the lab.

Remember that when you use the computers in the lab, you should work locally on the C: drive, but save your work to your home directory on the H: drive. The data for the labs should be loaded into C:\data. Since remotely sensed images are large, please do not copy any data to your H: directory unless absolutely necessary (and, if you do, you should first ZIP the files in order to reduce their size).

Introduction to IDRISI

The objective of this component of the lab is to introduce you to the IDRISI software package. You will be completing some of the "Using IDRISI" exercises using datasets provided by IDRISI (from Spain, and the US) in order to learn how to navigate the software, display layers and groups of files (Landsat band files, for example), discover how layers interact, create maps with both raster and vector data, and generate a 3D view and fly through (optional, but fun). You will be complete five of the Tutorial Part 1: Using IDRISI exercises and answer some questions (listed below) related to the tutorial exercises. You will then import some Landsat data of the Fraser Valley and answer some questions pertaining to that dataset.

Optional additional readings: There is a comprehensive IDRISI manual available ( C:\Program Files (x86)\IDRISI Selva\Documentation\IDRISI Manual.pdf) that you can copy onto a memory stick, should you like to explore IDRISI in more detail. (Note: you'll find the manual/tutorial files in C:\Program Files\IDRISI Selva\Documentation\ when working on the computers in Rm 114.)

LOAD DATA information: For this lab, and the succeeding labs, you will be using Windows Explorer to copy, paste, unzip and zip the data:

• Using Windows Explorer, go to: G:\courses\data\IDRISI_labs
• In this directory, you will see a subdirectory called Lab4, in that subdirectory there are two zip files. One contains the tutorial data used in Part 1 of this lab (Using IDRISI.zip), while the other contains the Fraser Valley Landsat images used in Part 2 of this lab (Landsat95.zip).
• In order to complete Part 1 of this lab, copy the Using IDRISI.zip file to C:\data -- these are large files, and you only need the files contained in the Using IDRISI.zip file to complete Part 1 of this lab.
• Double-click on the zip file (Using IDRISI.zip) and then drop / drag the Using Idrisi folder into C:\data.
• You should now see a Using IDRISI folder within C:\data (Make sure that you didn't end up with a Using Idrisi folder inside of an Using Idrisi folder--if so, move the data from inside the 2^nd folder into the 1^st folder.)

These folders contain large files, so

• Please do not save these folders / files to H: without zipping them first.
• You may not need to save your work if you answer the questions as you go. Do not save the zip file to H: if it is not necessary.

Assignment

Part 1: IDRISI Tutorial: Using IDRISI

• Launch IDRISI Selva from Start / All Programs / IDRISI Selva.
• Ignore any message that may appear about a "Data path not found." as we will address that problem below (Click on OK).
• We will have some hard copies of the relevant sections of the IDRISI tutorial available for you in the lab (a scanned copy of the relevant sections here). You can access the tutorial from C:\Program Files (x86)\IDRISI Selva\Documentation\IDRISI Tutorial.pdf (and copy it onto a memory stick, if you wish).

You will be completing the following exercises for this lab:

• 1.1 Using the IDRISI environment
• 1.2 Display: layers and groups files
• 1.3 Display: Layer Interaction Effects
• 1.4 Display Surfaces: Fly through and illumination (optional, but it is quick and fun!)
• 1.5 Display Navigation Map Query
• 1.6 Map Composition
• 1.7 Palettes, Symbols and Creating text layers (optional if you like creating maps)

Before you complete each exercise you should read through the "Specific Instructions and Questions" section below.

Specific Instructions and Questions for each Exercise

Exercise 1-1 The IDRISI environment

Read the first few pages about projects and working folders. Note that the project and working directories you will be using are different from those specified in the tutorial on page 6, Exercise 1-1 d) and e):

• Open IDRISI Explorer (either by clicking on File / IDRISI Explorer or by clicking on the [+] symbol that appears at the far left, immediately below the Idrisi toolbar.
• Click on Projects tab in the Idrisi Explorer window.
• Right click on the existing pathname (which might be C:\IDRISI Selva\Projects) and select Change projects folder.
• Navigate to C:\data, click on the Using Idrisi folder and then OK.
• Right click again on C:\data, select New project and again select the Using Idrisi folder.
• You have now set both the projects folder and the working directory. If you click on the Files tab you should see a list of the files stored in the Using Idrisi directory.
• There is no save icon (do not worry about this).

Question 1. Outline the differences between raster and vector data in terms of i) how the spatial data is stored and displayed, ii) how changing the scale of the composition affects the display of each type, and iii) how each type is suited for a particular analytical operation / application. (The differences between raster and vector data are discussed throughout the exercises, although Exercises 1-1 and 1-2 in particular cover the differences). (6)

Question 2: SIERRA234 is a composite of Landsat Bands 2, 3 and 4. What wavelengths do each band capture, and what feature characteristics does each band, in particular, highlight? (2)

Question 3. When you use the Ortho tool, you are creating an orthographic image. Define orthographic. (Note that this term differs from orthophoto.) (1)

Exercise 1-2 Display: Layers and Group files

Question 4. This dataset is from Spain. Use Google Earth and Google Maps (link) to familiarize yourself with the area. How far west from Madrid is the lake in the top of the image? (Use the shape of the lake to help identify the specific lake [it is actually a reservoir--could you deduce this from the satellite image?].) What is the name of the lake (use Google Maps)? (2)

Question 5: What does NDVI stand for? Why is it used (i.e., what is highlighted in NDVI images)? What is the formula (i.e., how are NDVI values calculated)? (3)

Question 6: Define, with examples, qualitative and quantitative data. What is ordinal, or ranked, data? (3)

Exercise 1-3 Display: Layer Interaction Effects

Question 7: Hand in the results of your Transparency exercise. To do this, use the Print option on the composer menu and ensure that you send your image to the appropriate colour printer. (2)

Question 8: SIERRA457 is a composite of Landsat Bands 4, 5 and 7. What wavelengths do bands 5 and 7 capture, and what feature characteristics do they, in particular, highlight? (2)

Since we don't have the Anaglyph glasses, skip over that element of the exercise.

Exercise 1-4: Display: Surfaces -- Fly through and Illumination

This exercise is optional, but it does demonstrate some of the 'neat' capabilities of Idrisi. (View the online tutorial. Note that the San Francisco data (Just for Fun ...) is not included in the Using Idrisi dataset provided in this lab.)

Exercise 1-5 Display: Navigation Map Query

Question 9: What are the BGR pixel values for the water in the SIERRA234 image? Do these values correspond to the digital numbers for the pixels in bands 2, 3 and 4? (You can determine this from the Feature Properties box.) What is the elevation of the lake? (2)

Exercise 1-6 Map Composition

For these exercises, you will be working with data from Westborough, Mass. USA.

If have have trouble finding some of your map features you are told to create, try resizing/expanding your window.

Important! When going through Exercise 1-6 f), you must add a subtitle to the map that contains your name and lab section #. (Select a font size of 10.)

Question 10. Print the map composition you created in this exercise. (4)

Exercise 1-7 Palettes, Symbols and Creating text layers

Exercise 1-7 is optional.

Part 2: Using Landsat data of the Fraser Valley

For this part of the lab you will take what you learned in the tutorials, and apply it to a local dataset: seven bands of Landsat data, obtained in 1995, of the Fraser Valley around Abbotsford and Chilliwack (map). A vector file of some roads in this area is also provided.

Loading the data

• Using Windows Explorer, go to: G:\courses\data\IDRISI_labs\Lab4.
• In order to complete Part 2 of this lab, copy the Landsat95.zip file from Lab4 to C:\data.
• Double-click on the zip file (Landsat95.zip) and then drop / drag the Landsat95 folder into C:\data.
• You should now see a Landsat95 folder within C:\data.

Start IDRISI

• Launch IDRISI Selva from Start / All Programs / IDRISI Selva.
• Click on Projects tab in the Idrisi Explorer window.
• Right click on the existing projects folder and select Change projects folder.
• Navigate to C:\data, click on the Landsat95 folder and then OK.
• Right click again on C:\data, select New project and again select the Landsat95 folder.
• You have now set both the projects folder and the working directory. However, if you click on the Files tab nothing should be listed--this is because the data is not yet in an IDRISI format (i.e., we need to import the data before IDIRISI can recognize it).

Data conversion:

One of the first steps you must complete before starting most Digital Image Analysis (DIA) projects is to import the data (i.e., converting the data from its 'native' format into a software-specific format). For Part 2 of this lab you have been given two different types of data--some raster remotely sensed data (Landsat data), and some vector data (in this case some ESRI ArcGIS data). Thus, your first step will be to use the IDRISI Import command to import the raster Landsat data and the vector GIS data into the IDRISI working directory. Click on File / Import and you can see all of the options for importing data into IDRISI. You will first import the Landsat data.

• Click on File / Import
• Select the option for Government Data Provider Formats - Landsat ETM
• In the popup menu,
  • Select GeoTIFF as the Data format type.
  • Starting with Band 1: input GeoTIFF: click on the … button and select Landsat95b1. Accept the default IDRISI name for the output file.
  • Repeat this process for the 6 remaining bands. Click on the check box , and then click on the .... button to open up the dialog window in order to select the next band. Ensure that the output file names are spelled the same, but end with the appropriate band #. (Although this is not necessary, it is good practice to name files so that their relations are obvious). (Note that we don't have Band 6L or 6H data). (Since the first Output Idrisi image filename was converted to all uppercase, it would be better if you retyped that filename to lower case in order to follow the case of the other filenames.)
  • Click on OK.
  • It will take IDRISI a minute or so to process the files (it imports them one at a time). After some time you should see an image of the the first band appear on your screen.
  • In the Explorer window, click on Files and you should see all 7 bands listed. (NOTE: It sometimes takes a few minutes for Explorer window to refresh and for the new files appear in the listing.)
• You should next import the roads vector file that is in ESRI ArcGIS shapefile format (this dataset was extracted from a DMTI dataset that contained all of the roads of BC).
• Click on File / Import
• Select the option for Software Specific Formats - ESRI formats - Shapeidr
  • In the popup menu, ensure that the Shapefile to IDRISI button is checked.
  • Browse for the Roads file (it should be the only choice).
  • Call the output IDRISI vector file Roads.
  • In the Reference system, change the reference from Plane (basically an undefined planar coordinate system) to UTM-10N.REF (The Landsat images have been georeferenced to a UTM [zone 10, North] coordinate system, so if we assign the same georeference system to the roads [knowing that roads had previously been projected to a UTM coordinate system within ArcGIS], the roads and the Landsat images will be geographically referenced together.)
• To check that the two import processes worked as expected:
  • Close down all of the IDRISI display windows
  • Open the Display Launcher
  • Add the Landsat band1 raster layer using the grayscale option
  • In Composer, select Add layer, Vector, select Roads, and specify Outline white.
  • The roads should appear overlaid on the Landsat image.
• Zoom in and out of the image(Page Up / Page Down), and use the arrow keys to pan, to help you locate Hope, Harrison and Chilliwack in the image (using Google maps may also help).
• Close your IDRISI display windows.

Displaying the data and creating images

The first thing you will do is to create a True Colour composite image. To do this, you need to display band1 in blue, band2 in green and band3 in red.

Note: The naming convention for composite images is variable. While many people place the band numbers following a short-to-long wavelengths convention (e.g., Landsat123 corresponds to blue - green - red [the bands presented in increasing wavelength]), others use a naming convention that follows the computer primary colour convention (RGB), in which case the file would be named Landsat321 (where band 3 is assigned to Red, band 2 to Green, and band 1 to Blue). In the IDRISI tutorial the first convention (ordered by increasing wavelength) was followed (as discussed in footnote 19 on page 27 of the Tutorial), so we shall follow that convention in our labs.

• To create a true colour composite of bands 1, 2 and 3:
  • Under Display, select Composite.
  • Add the three imagebands (band1 as blue, band2 as green, and band3 as red).
  • Name the output file Landsat123.
  • Accept the defaults for Contrast stretch type and Output type, but check the box beside Omit zeros ....
  • Add a title: Fraser Valley, 1995 [True colour image of Landsat bands 1, 2, 3]
• After the composite image appears, close the window and re-display the composite image using the Display Launcher (so that the title appears at the top of the window).
• Add the vector layer roads as above (recall: pressing v brings up the add a vector layer dialog window).
• Zoom into the western part of the image (somewhere just north of Chilliwack, ensuring that Hope is still visible).
• Set the map scale (RF) to 300,000 (omitting the comma when specifying the scale; right-mouse clicking within the image allows you to specify the map scale).
• Add text labels for Hope, Harrison and Chilliwack (you can add more if you wish):
  • In the Tools menu, click on the Digitize tool (looks like a bulls eye) .
  • Name the layer to be created Cities.
  • Select Text as the Layer Type.
  • Select IDRTEXT as the Symbol file for display (do this after specifying text as the layer type).
  • Accept the other defaults, and click on OK.
  • Click in the image window where you want the beginning of the text to go.
  • Another window will appear, in which you enter the name of the city (Caption).
  • Select the caption position relative to the anchor point (e.g., should the text appear above and to the right of where you clicked in the image, or below and to the left?).
  • Specify the rotation angle for the text (the default Rotation Angle of 90.0 results in the text flowing horizontally).
  • Repeat for each city; you can add more names if you like (e.g., Cultus Lake).
  • Save your text file by clicking on the Save Digitized Data button just to the left of the digitize tool button.
• Click on Map Properties
  • Add a north arrow and a scale bar.
• Right click on the title
  • Using the caption, add your name, lab# and date to the composition.
• To save your image, in the Map Composition window, click on Save
  • Pick the Save Composition to Map file option.
  • This will save your composition for further editing.
• If you are finished, print your map composition using the Print command in the Map Composer. Review the material presented in Exercise 1-6 if you need a reminder on all of the specific instructions on preparing and printing a map composition.

Question 11. Hand in the printout of your Landsat123 image composition. (4)

Question 12. The roads, unfortunately, don't line up with the Landsat images (although both use the UTM coordinate system, the georeferencing of the Landsat images resulted in some positional error being introduced). Zoom into the area just southeast of Chilliwack where Hwy #1 runs more or less due east-west, and use the Measure tool to determine how far off (in a north-south direction only) the vector roads are from the roads as imaged in the Landsat images. (1)

Question 13. Why do Cultus Lake and Harrison Lake appear blue in this image, but not the Fraser River? (1)

• Repeat the Composite process with Landsat bands 2, 3 and 4, assigning band 4 to Red, band 3 to Green and band 2 to Blue, and naming the output file Landsat234. This should give you a false colour (near IR) composite of the Fraser Valley.

Question 14. Hand in the printout of your Landsat234 composite image. Include the same information as in the previous map, but provide an appropriate title. (2)

Question 15. About 6 km southeast of the intersection of Hwy #1 and Hwy #9 are some (non-natural) features that dot the landscape (generally extending north-south; they occur throughout the Chilliwack River valley. What are they? (Provide your reasoning.) Use both Landsat123 and Landsat234 to help you in your identification. (2)

• with Display Launcher, bring up band 6
  • Select grayscale
  • Click on Layer properties and alter the min and max values of contrast to get a better contrasting image (I found that raising the min value alone does the trick).

Question 16. Print your band 6 image. (1)

Question 17. What wavelengths are captured by Landsat band 6? What information is being presented in this band's image, and how could it be used? (3)

Question 18. Is the spatial resolution of this image different from that of the other bands? Has it been resampled (if so, how can you tell)? (Hint: If you create a Raster Group [in Idrisi Explorer / File, highlight all of the Landsat images, right-mouse click and select Create > Raster Group] it will make it much easier to determine this. From within the Raster Group, display Landsat 6 and Landsat123. Then, click on the Group Link button and zoom into the band 6 image; you should then also zoom into the same area in the Landsat123 image. The Group Linked Zoom instructions in Exercise 1-5 provide more details, if necessary.) (3)

Question 19. Describe the strengths of each of the three images (Landsat123, Landsat234, band 6) you have explored above in interpreting the landscape. (6)