LIST OF EXERCISES

Exercise No. Title

1 Introduction of Digital Image and Image Processing Tool (ENVI)

2 Image Pre-processing
3 Image Enhancement
4 Image Registration
5 Band Math Operation
6 Pixel-based Unsupervised Classification
7 Pixel-based Supervised Classification
8 Post Classification
9 Accuracy Assessment
10 Map Analysis
11 Decision Tree
2
 3

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 1

Introduction of Digital Image and Image Processing Tool (ENVI)

Title

I. Introduction:

Digital images are electronic snapshots taken of a scene or scanned from documents,
such as photographs, manuscripts, printed texts, and artwork. It is sampled and mapped
as a grid of dots called picture elements (or pixels). Each pixel is assigned a tonal value
(black, white, shades of gray or color) which is called digital numbers (DN). The binary
digits (or bits) for each pixel are stored in a sequence by a computer and often reduced to
a mathematical representation. The bits are then interpreted and read by the computer to
produce an analog version for display or printing. Digital image are processed and
interpreted thru image processing software like ENVI or Environment for Visualizing
Images.
ENVI is the software for the visualization, analysis, and presentation of all types of
digital imagery from multispectral to hyperspectral datasets. It can perform spectral
analysis, geometric correction, terrain analysis, radar analysis, and has raster and vector
capabilities from various image sources. It can also perform image enhancement and
image analysis, among others.

II. Objectives
At the end of this exercise student will be able to:
1. Explain the characteristics of digital image, multispectral image, and false color
image
2. View digital image in different band combinations/composition in ENVI software
3. Name common digital image processing software

III. Materials/Data
Sample dataset: Landsat5 TM
IV. Procedure/Preparation:

Digital image of Pathumthani, Thailand acquired from Landsat Enhanced Thematic

Mapper (Landsat ETM) data will be used and analyzed in this exercise. You will open and
manipulate the individual bands of this image using ENVI software. The series of
instructions below are designed to be self-sufficient however; questions are welcome if
confusions may arise. You will now explore the ENVI software package.
 4

1. Getting started with Basic Tool (ENVI):

1.1. Run ENVI.

1.2. The MAIN menu bar appears (as in figure 1) with options such as “File”, “Basic
Tools”, “Classification”, etc. This bar can be oriented horizontally or vertically on
screen by modifying the “Preferences” under “File” Menu.

Figure 1. ENVI Main Menu

2. Viewing/Displaying Digital Image

2.1. To open an image file, select File Open Image File. Then navigate the above-
mentioned file in “Materials/Data”, and click Open.
2.2. The “Available Band List” (figure 2) dialog will appear on your screen. This list
allows you to select spectral bands for display and processing. Note that you
have the choice of loading either a grayscale or an RGB color image.
2.3. To open gray scale image, select Gray Scale option and Select one of the bands
listed at the top of the dialog by clicking on the band with the left mouse button.
The band you have chosen is displayed in the field marked “Selected Band”.
Click “Load Band” to load the image into a new display with this band.
2.4. To open RGB image, select RGB Color option and click the band you wish to
display as Red (R), Green (G), and Blue (B). Click “Load RGB” to load the image
into a new display with the 3 bands you just selected. (Refer to Landsat5 TM
spectral characteristics.)

Landsat 5TM bands Characteristic

Channel Wavelength Range (µm)
TM 1 0.45 - 0.52 (blue)
TM 2 0.52 - 0.60 (green)
TM 3 0.63 - 0.69 (red)
TM 4 0.76 - 0.90 (near IR)
TM 5 1.55 - 1.75 (short wave IR)

2.5. ENVI will displays image in three (3) different image windows: a Main Image
window, a Zoom window, and a Scroll window.
2.6. To try different combinations of bands to be open in RGB or color composite,
simply change the R, G, and B color gun to your choice image band.
2.7. To display a pseudo color image, just follow the steps 2.1 to 2.3, then click Tools
 Color Mapping  ENVI Color Tables (figure 3) from “Image window”, then
select the color you want from color tables (say Rainbow).
 5

Guide Questions:
1. What is the projection, datum, and pixel size of the image?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
2. How many samples, and lines does an image have?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
3. What is the mean/average Digital number for each bands?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
4. What have you observe with these three windows in terms of extent?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

5. What is natural color composite? False color composite? Pseudo color?

________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

6. Displaying Cursor Location and Image Profiles

3.1. To display the cursor location and value, double click on the Main window or
select Tools -> Cursor Location/Value from the Main window. A dialog box will
appear displaying the location of the cursor in the Main, Scroll, or Zoom windows.
 6

The dialog also displays the screen value (color) and the actual data value of the
pixel underneath the cross-hair cursor.
3.2. To exactly position the cursor, select Tools -> Pixel Locator from the Main
window. A dialog box will appear displaying the x and y location of the cursor in
the Main, Scroll, or Zoom windows. To change the cursor position, change the x
and y values appropriately and “Apply”.
3.3. X (horizontal), Y (vertical), and Z (spectral) profile plots can be selected from
Select Tools -> Profiles and displayed interactively. Change the locations of the
profiles by moving the red box in the image window.
a. Select X profile. A window will appear that contains a horizontal profile. In
the image window, a red line appears horizontally across the image. This
red line of the profile for which the DN numbers are plotted.
b. Select Y profile. A window will appear labeled “Vertical profile’. In the mage
window, a red line appears across the image. This red line is the line of the
profile for which the DN numbers are plotted.
c. Select Z profile. This gives you the DN values for the bands in the image
for the pixel on which the red box the image window is centered. The Z
profile gives you the spectral signature of the pixel of interest. You can
change the wavelength region by selecting “Edit” and then “Plot
parameters”. Enter 2.5 for the maximum wavelength and hit “Apply”.

Figure 2. a) Available Band List b) Color Mapping Window

Guide Questions:
1. Describe the z-profile of the water. What part/band it became maximum?
Minimum?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
 7

________________________________________________________________

2. Describe the z-profile of the vegetation. What part/band it became maximum?

Minimum?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

3. Describe the z-profile of the bareland. What part/band it became maximum?

Minimum?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
V. Results and Discussion

VI. List of References

1. ENVI Tutorial
2. https://www.library.cornell.edu/preservation/tutorial/intro/intro-01.html
8
 9

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 2

Image Pre-processing
Title

I. Introduction:

Image pre-processing are preliminary activities perform before an image is processed

and analyzed. This includes layer stacking, image subset, radiometric correction, and
image enhancement. This step is necessary for some reasons such as to improve of the
image data by minimizing unwanted distortions or to enhance some image features
important for further processing, or to accelerate the further image processing.
In this exercise, layer stacking, image subset, and saving image, while image
enhancement will be performed in the next exercise.

II. Objectives

At the end of this exercise student will be able to:

1. Perform layer stacking, image subset, and save image

2. Perform image enhancement

III. Materials/Data
Sample dataset: Landsat5 TM

IV. Procedure/Preparation:

Digital image of Philippines acquired from Landsat5 Thematic Mapper (Landsat TM) data
will be used in this exercise. You will perform layer stacking, image subset, save image,
and perform various techniques of image enhancement.

1. Layer Stacking:
Layer Stacking is employ to build a new multiband file from georeferenced images of
various pixel sizes, extents, and projections. The input bands will be resampled and re-
projected to a common user-selected output projection and pixel size. The output file will
have a geographic extent that either encompasses all of the input file extents or
 10

encompasses only the data extent where all of the files overlap. To perform layer stacking,
the following steps are perform:

a. Select one of the following options from the ENVI main menu bar:
 Basic Tools  Layer Stacking
 Map  Layer Stacking
The Layer Stacking Parameters dialog appears.

b. Click Import File. The Input File dialog appears.

c. Select an input file and perform optional Spatial Subsetting and/or Spectral
Subsetting, then click OK. The Layer Stacking Parameters dialog appears. ENVI adds
the input file to the Selected Files for Layer Stacking list.

Figure 3. Layer Stacking Window

d. Repeat the file selection using the Import File button for each input file to include in
the new output file. Or you can you can hold simultaneously with “Shift” key.
e. Select Inclusive or Exclusive to specify the output file range. Inclusive creates an
output file with a geographic extent that encompasses all the input file extents while
Exclusive creates an output file that contains only the data extent where all the files
overlap.
f. Select output to File or Memory to save the multiband image.
g. Select an output map projection from the list.
h. Enter the X Pixel Size and Y Pixel Size. Enter the pixel sizes in the units selected
under the map projection section.
i. From the Resampling drop-down list, select a resampling method. Nearest Neighbor
uses the nearest pixel without any interpolation to create the warped image. Bilinear
performs a linear interpolation using four pixels to resample the warped image. Cubic
Convolution uses 16 pixels to approximate the sine function using cubic polynomials
to resample the image. Cubic convolution resampling is significantly slower than the
other methods, however.
j. Click OK. ENVI then adds the resulting output to the Available Bands List.
 11

Guide Questions:
1. How the processed (multiband) dataset does appears in the available bandlist?
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________

2. Subsetting Data
Image subset is perform before image processing to minimize the size of the image
thus save computer memory. This can be categorized as spatial, spectral, or statictics
subset. Perform spatial subset by following the steps:
a. Click Basic Tools  Resize Data (Spectral/spatial). Then in the Select Spatial
Subset dialog, select the image to subset by from the Subset by Image drop-down
list.
b. Enter the starting (1000) and ending (1000) values of the Samples (NS), and 2000
x 2000 for Lines (NL).
c. Click OK.

d. Perform image subset using “image’ and “map”.

e. Creating ROI (region of interest) and Image Subset using ROI.
 Using the same image (in procedure A), create your ROI (of not less than
1000 x 1000 pixels) using ROI Tool: Tools  Region of Interest  ROI
Tool. In the “ROI Tool” window, choose polygon as the ROI Type, then
draw the polygon on the scroll window.
 You may save the ROI for later use.
 12

 Now resize the image using the created ROI: Basic Tools Subset via
ROIs. You may select the bands that you want to subset with this ROI. You
may save the resized image.
 The resized image will then be displayed in the available band list.

Guide Questions:
7. Describe the meaning of 1000 x 1000 and the 2000 x 2000 in the NS/NL field?
Try 3500 x 2000, and 5000 x 10,000. Describe the upper-left and lower-right
grid coordinates of these images.
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

8. Which of these techniques (samples/lines, image, map) allows you more full
control in image subset?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

9. Does the color of the spatial subset image change? How about the pixel size?
Image size?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
f. Perform spectral subset (bands 2, 3, and 4) by selecting “spectral” subset instead
of “spatial”.

1. What image characteristics (size, pixel size, extent, number of bands) do

change after subset?
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

2. What is the difference between spatial and spectral subset?

________________________________________________________________
 13

________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

3. Saving Image

Save the output in procedure “a” and “b”:

1. In the Main Menu, click File  File Save as
2. Or in the Image window, click Save Image As  Image File

Guide Questions:
4. What are the various image or file format that can be save in ENVI? Write their
respective characteristics accordingly.
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________
________________________________________________________________

V. Results and Discussion

VI. List of References

1. ENVI Tutorial
2. ENVI Help
 14

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 3

Image Enhancement
Title

I. Introduction:

Image enhancement is a pre-processing activity performed to a satellite image in

order to improve its visual appearance thus making the analysis easier, faster, and
reliable for the analyst. The most common types of image enhancement tools can be
found in most GIS and image processing software. These are contrast enhancement,
linear stretch, histogram equalization, density slicing, and edge enhancement. The
choice of method is dependent on the objectives of the analysis.

In this exercise, linear stretch method will be performed.

II. Objectives

At the end of this exercise student will be able to:

1. Perform image enhancement thru various interactive stretching techniques

2. Discuss the concept of image enhancement (interactive stretching)

III. Materials/Data
Sample dataset: Landsat5 TM

IV. Procedure/Preparation:

1. Enhancement by Histogram Stretching

a. Use the output image (from exercise no. 2) in this exercise. Run ENVI and Open
an RGB Image File in True Color composite. Enhance this image using different
image enhancement technique of “Interactive Stretching”.
b. From the Display menu bar, select “Enhance>Interactive Stretching”.
 An Input and an Output histogram appear in the Interactive Contrast Stretching
dialog. They show the current input data and applied stretch respectively.
 15

 Two vertical dotted lines mark the current minimum and maximum values of the
stretch. For colored images, the color of the histogram matches the color of the
selected band (the red band is displayed by default). At the bottom of the
Interactive Histogram window, the stretch type and the histogram source are
listed.

c. Select “Histogram_Source” from Interactive Stretch window.

 Input histogram information can be extracted from the Main Image window, from
the Scroll window (subsampled), from the Zoom window, from the entire image
band (all pixels of the data), or from a user defined region-of-interest (ROI).

2. Enhancement by Linear Stretch Type, Changing Minimum and Maximum

Stretch value

a. From within the Interactive Contrast Stretching dialog, select

“Stretch_Type>Linear Contrast Stretch”.
b. To change the minimum and maximum stretch values, enter the values in the text
boxes at the top of the dialog or a percentage of the data (e.g., 2% for the
minimum and 98% for the maximum).
c. Or Click with the left mouse button on either of the dotted stretch bars and drag
them to a new location. If Stretch bar was locked, you could unlock it by Select
“Options>Stretch bar Unlocked”.
d. Click Apply to apply the stretch to the displayed data.
e. To re-display the original stretch select “Options>Reset Stretch”.

3. Enhancement by Piecewise Linear Type

a. Select “Stretch_Type>Piecewise Stretch”.

 16

b. Click the middle mouse button anywhere in the Input Histogram plot to add a node
to the transfer function. Line segments are plotted to connect the endpoints to the
plotted node symbol.
c. To move the position of a point, click the left mouse button on the symbol and
drag it to a new position.
d. To delete points, click on the symbol with the right mouse button.
e. Click Apply to apply the stretch to the displayed data.

4. Equalization Stretch Type

To automatically scale the data to equalize the number of DNs in each histogram bin:
a. Select “Stretch_Type>Equalization”. The input histogram shows the unmodified
data distribution. The output histogram shows the equalization function as a red
curve, and the stretched data distribution is shown superimposed in white.
b. Click Apply to apply the stretch to the displayed data.

Guide Questions:
1. Does performing image enhancement (linear stretching) affect the quality (pixel
value) of the data?
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________

2. Discuss the effect of each method above to the original image.

__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________

V. Results and Discussion

VI. List of References

3. ENVI Tutorial
4. ENVI Help
5. http://www.seos-project.eu/modules/remotesensing/remotesensing-c05-
p02.html
6. https://www.e-education.psu.edu/geog883/node/557
 17

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 4

Image Registration
Title

I. Introduction:

Image registration or geometrical correction is a process to reference images to

geographic coordinates or contents on the image (or map) will be assigned a spatial
coordinates. It is an important prerequisite which must be performed prior to using images
in geographic information systems (GIS) and other image processing programs. To
process the data with other data or maps in a GIS, all of the data must have the same
reference system. The term is also known as geo-referencing.
In geo-referencing, image points and pass points need to be searched, which then
can be recognized in the coordinates. Pass points or ground control points (GCP) are
usually determined with a GPS receiver on the terrain or with maps. Visual street
crossings, bridges over water, etc. can be identified, and their coordinates will be noted.
These points will then be coordinated with identical image points of the not yet geo-
referenced satellite image. These correlations can ensure projections with the help of
various additional procedures.
In ENVI, there are two methods to register an image or map; Image-to-Map and
Image-to-Image. Image-to-Image requires to display both the base (reference) image and
the warp (un-registered or un-corrected geometrically) image, while image-to-map
requires only the warp image to display then coordinates of GCPs are encoded or entered
manually. In this exercise, both will be performed.

II. Objectives

At the end of this exercise student will be able to:

3. Perform image registration in ENVI with acceptable accuracy

4. Discuss the concept of image registration

III. Materials/Data
Aerial image of VSU – for image-to-image registration
Topographic scanned maps – for image-to-map registration
 18

IV. Procedure/Preparation:

1. Image-to-Map (Registering a scanned Topographic Map)

5. Run ENVI.
6. Open the image (51373_er_crop.tiff or 51374_er_crop.tiff) to be registered as true
color.
7. Geometric Correction by Image-to-map registration.
 From Main menu, choose Map > Registration > Select GCPs: Image to Map.
 Choose the image to register to the map by clicking on the display name in dialog
(in case of multiple image are displayed).

8. In the Select Registration Projection list, select the output projection.

 Select UTM projection for output projection.
 Select WGS84 for Datum and Zone 47 N.
 Enter 30.0 for X Pixel size and Y pixel Size.
 Click Ok.

Guide Questions:
1. Why UTM projection, WGS84 datum, and zone 47N must be chosen? What are
the bases in selecting appropriate projection, datum and zone number?
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________

2. What is the unit of 30.0? What does it refers to?

___________________________________________________________________
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________

9. Collecting Ground Control Points.

 In the Main Image window, position the Zoom box over the desired GCP area.
 In the zoom window, click the left mouse button on a specific pixel to position the
cursor over that pixel or portion of that pixel.
 The coordinates of the selected location appear in the Ground Control Points
Selection dialog in the fields labeled Image X and Image Y. Subpixel coordinates
are supplied to provide higher accuracy in selecting GCPs.

10. Select common reference features in both the map and the image
 Enter the map coordinates, enter the easting and northing (or lat/long) of the
selected GCPs into E and N text boxes, respectively.
 19

 Once the desired pixel is selected in the image and the map coordinates have
been entered, click Add Point in the Ground Control Points Selection dialog to
add the point to the list of GCPs.
 To view GCP List, click Show List.
 When a GCP is added, a marker is placed in the image. The marker indicates the
selected pixel (or subpixel location); the center of the marker (located under the
cross-hair) indicates the actual GCP location.
11. Add additional GCPs using the same procedure.
 After a number of GCPs sufficient enough to conduct a 1st degree polynomial warp
have been selected, the total RMS error is displayed in the Ground Control Points
Selection dialog and the RMS error for each point is listed in the GCP List table.

12. Minimizing RMS Error

 To see which points have the highest RMS errors, select Options>Order Points
by Error in the Ground Control Points Selection dialog.
 To have GCPs selectively ignored during registration, click on the point to be
ignored and click the On/Off button.
 To turn the GCP on again, click on the point in the list and click the On/Off button
again.
 Accepted minimum RMS Error is less than one pixel.
Guide Questions:
1. Why do selection of points must be done in “zoom” window?
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________

2. If the RMSE (random mean standard error) is 35.5, is this acceptable or not? Why
or why not?
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________

13. Warping and Resampling

 In the Ground Control Points Selection dialog, select Options>Warping File

 When the Registration Parameters dialog appears, select the warping method
from the Warp Method button menu.
 From the Resampling button menu, select the resampling method.
 Set the Background Value to be 255.
 Select output to File and Enter output Filename and Click Ok.
Guide Questions:
1. What is the Background Value of 255 mean? What if 0 is set instead? What happen
 20

to the image?
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________
___________________________________________________________________

Follow all the steps above to register the second topographic map.
2. Image-to-Image Registration

a. Geometric Correction by Image-to-image registration.

 Both base image and warp image must be first displayed.
 From Main menu, choose Map > Registration > Select GCPs: Image to Image.
 The Image to Image Registration dialog appears. In the Base Image list, select
the display group corresponding to the base (reference) image. In the Warp Image
list, select the display group corresponding to the warp image.

 Click OK.
 The Ground Control Points Selection dialog appears.

Guide Questions:
1. Write below the name of image in your respective display.

Display #1: ______________________________________________________

Display #2: ______________________________________________________
Base image file name: _____________________________________________
Warp image file name: _____________________________________________

b. Collecting Ground Control Points (Image-to-Image)

Before performing registration, it is advised to rectangular subset both images of the

same extent.

Identify GCPs in the two displayed images by locating pixels in the Zoom windows, as
the steps below describe. Pixel information is loaded into the Ground Points Selection
dialog. Whole-number pixel coordinates correspond to the upper-left corner of the pixel.
The x,y values increase to the right and bottom of the pixel, respectively.
 21

The pixel fraction available in the Zoom window is proportional to the zoom factor. For
example, at a zoom factor of 4x, the pixels are divided into four subareas. At a zoom factor
of 10x, positioning is possible to 1/10th of a pixel. The GCP marker is placed according to
the subpixel position in the Zoom window. Subpixel locations are supplied to provide
higher accuracy in selecting GCPs.
 In the Image window of each image, position the Zoom box over the desired GCP
area.
 Add individual GCPs by positioning the cursor in the two images to the same
ground location. Examine the locations in the two Zoom windows, and adjust the
locations as needed by left-clicking in each Zoom window. Subpixel positioning is
supported in the Zoom windows. The larger the zoom factor, the finer the
positioning.
 The sample and line coordinates (in both images) appear in the Ground Control
Points Selection dialog, in the Base X, Y and Warp X, Y fields, respectively.
Subpixel coordinates are shown as floating-point values.
 In the Ground Control Points Selection dialog, click Add Point to add the GCP to
the GCP list.

 To view the list of GCPs, click Show List. The Image to Image GCP List appears
with the GCPs listed in a table. For a description of the GCP List, see Using the
Image to Image GCP List.
 When the GCPs are added to the list, a marker is placed in the Image windows of
both the base and warp images. The GCP marker consists of an ID number next
to an encircled crosshair. The marker indicates the selected pixel (or subpixel
location). The center of the marker (located under the crosshair) indicates the
actual GCP location.
 Add additional GCPs by following the same procedure.
 When you select at least four GCPs, the predicted x,y coordinates for the selected
warp, the x and y error, and the RMS error are listed in the Image to Image GCP
List table.

c. Add additional GCPs using the same procedure.

 After a number of GCPs sufficient enough to conduct a 1st degree polynomial warp
have been selected, the total RMS error is displayed in the Ground Control Points
Selection dialog and the RMS error for each point is listed in the GCP List table.

d. Minimizing RMS Error

 To see which points have the highest RMS errors, select Options>Order Points
by Error in the Ground Control Points Selection dialog.
 To have GCPs selectively ignored during registration, click on the point to be
ignored and click the On/Off button.
 To turn the GCP on again, click on the point in the list and click the On/Off button
again.
 Accepted minimum RMS Error is less than one pixel.
 22

Guide Questions:
1. Fill-up below the coordinates of the warp and base image accordingly:

Image X Y X Y Residual
Warp
Base
RMSE =

e. Warping and Resampling

In the Ground Control Points Selection dialog, select Options>Warping File
When the Registration Parameters dialog appears, select the warping method
from the Warp Method button menu.
 From the Resampling button menu, select the resampling method.
 Set the Background Value (the DN value used to fill areas where no image data
appears in the warped image) by entering the DN value in the Background Value
text box.
Select output to File and Enter output Filename and Click Ok.

Guide Questions:
3. Display both the registered and the un-registered images from procedure number
1 and 2, then link them accordingly, and display cursor value. Write your
observation below.
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________
__________________________________________________________________

V. Results and Discussion

VI. List of References

7. ENVI Tutorial
8. ENVI Help
9. http://www.seos-project.eu/modules/remotesensing/remotesensing-c05-
p02.html
10. https://www.e-education.psu.edu/geog883/node/557
 23

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 5

Band Math Operation

Title

I. Introduction:

Band Math is a flexible image processing tool with many capabilities not available in
any other image processing system. Its function accesses data spatially by mapping
variables to bands or files. In the Band Math dialog, users can define bands or files used
as inputs to call a user band math function, and result is written to a file or memory.
Example of math operation are sum, average, multiplication, division, band ratios and
other arithmetic and/or logical operations. This tool is very important especially in doing
analysis.
Writing band math expressions has four basic requirements. It must be in single IDL
(interactive data language); all bands used must have the same dimensions; all variables
must be Bn or bn (e.g. B1, B2, …, Bn or b1, b2, …, bn); and result must be in the same
dimension as of the input bands.
In this exercise, band ratio (i.e. NDVI) and logical statement will be performed.

II. Objectives

At the end of this exercise student will be able to:

1. Use band math tool in ENVI to perform mathematical/logical operations

2. Explain band ratios such as NDVI (normalized differential vegetation index), EVI
(enhanced vegetation index), among others
3. Explain the various data type of imagery

III. Materials/Data
Landsat 7 (L7_IN_2005)

IV. Procedure/Preparation:

1. Familiarization of Data Type

 24

Image dataset can be byte, integer, long integer, floating-point, double-precision,

signed integer, unsigned integer, etc. Each data type occupies a certain space in computer
memory for example, byte occupies one byte per pixel, floating-point occupies four bytes,
and double precision occupies eight bytes. So an image of 100,000 x 100,000 pixels will
occupy around 37 Gigabytes, computer might crashed. Also, knowing data type especially
for the output image is essential in order to avoid loss of data. For example, a DEM image
with data type of double-precision floating-point will be downgraded to “integer” for
example. This could save a lot of memory however, an elevation of 451.5 would be
converted as 451, and 6556 would be read as 1, since the maximum value for integer is
6555 only.
To know the data type of an image, simply click the name of the image in the available
band list window, the type of data is indicated below. The image “bhtmref.img” below is
byte type.

Guide Questions:
1. Below are sample mathematical operations, indicate the final answer of each
expression, data type, and space on the memory it would occupy:

Expressions Answer Size/Memory Space

1. 8 + 10
2. 2 + 3.6
3. 5 + 6 * 4
4. (5 + 3) * 2.5
5. Float (4) + 10/3
6. 5 + 10/float(3)
 25

2. Making NDVI Image in Band Math

a. Subset file only 2000 x 3000 pixels (NS: 4501-6500; NL: 4001-7000) of the
“L7_IN_2005” and derive NDVI image using Band Math Operation
b. Select Basic Tools>Band Math from Main menu bar. The Band Math dialog
appears.
c. Enter the IDL expression you wish to have evaluated in the “Enter an
expression” text box. Use variables in place of band names or file names (the
variables will be assigned in the next step). Variable names must begin with
the character "b" or "B" followed by up to 5 numeric characters.
d. An infrared image band near 0.8 µm should be used for the b1 variable while
a red band near 0.6 µm should be used for the b2 variable followed by NDVI
formula and Landsat 7 ETM+ wavelength.

NIR  Re d
NDVI Formula =
NIR  Re d
Table 1. Landsat 7 ETM+ bands Characteristic
Channel Wavelength Range (μm)
TM 1 0.45 - 0.52 (blue)
TM 2 0.52 - 0.60 (green)
TM 3 0.63 - 0.69 (red)
TM 4 0.76 - 0.90 (near IR)
TM 5 1.55 - 1.75 (short wave IR)
TM 6 10.4 - 12.5 (thermal IR)
TM 7 2.08 - 2.35 (short wave IR)

e. (float(b1) - b2) / (float(b1) + b2) could be entered into the text box. Two
variables are used in this expression — b1 is one variable for NIR band, b2 a
second variable for Red band.
f. After a valid expression is entered, click OK. The Variable to Band Pairings
dialog appears.
g. In the Variables to Bands Pairings dialog click on the variable B1 in the
Variables used in expression text box. Then click on the NIR band in the list
labeled Available Bands List. Continue to assign a B2 variable by Red band
from the list.
h. Select Result to File and enter the filename then Click Ok

Guide Questions:
3. Examine the basic statistics of the NDVI image, write the answer below:
 26

Minimum value: _________________

Maximum value: _________________
Mean: _________________
Standard deviation: _________________
Number of bands: _________________
Area in has. of the whole image: _________________

V. Results and Discussion

VI. List of References

11. ENVI Tutorial

12. ENVI Help
13. https://www.harrisgeospatial.com/docs/bandmath.html
 27

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 6

Pixel-based Unsupervised Classification

Title

I. Introduction:

Digital image classification techniques group pixels to represent land cover features.
Land cover could be forested, urban, agricultural and other types of features.
Classification can be pixel-based or object-based approach. In pixel-based, pixels are
group according to their values or digital numbers while in object-based, an image is
classified by vectors such as lines and arcs that define a feature on the image. Since pixel-
based is based on the brightness value thus it is most preferable for low to moderate
resolution images, while object-based is suitable for high resolution.

Image classification can be performed with (supervised) or without (unsupervised)

analyst’s intervention in grouping of pixels. In this exercise, unsupervised technique will
be employed wherein computer clusters the n-dimensional space with best separation
then, analyst assign the class names of the groups/clusters.

II. Objectives

At the end of this exercise student will be able to:

4. Define image classification

5. Discuss pixel-based and object-based classification
6. Perform unsupervised image classification

III. Materials/Data
Landsat 7 (L7_IN_2005)

IV. Procedure/Preparation:

1. Perform image classification, subset image such that the upper-left and lower-right
(pixel) coordinates to be 4200, 3600 (samples) and 6500, 6500 (lines), respectively.
 28

2. Original image may be closed. Perform unsupervised classification of the resized

image.
3. In the menu bar click File  Basic Tools  Classification
4. Or Go to "Classification" in the main ENVI menu bar.
 Select "Unsupervised". Choose "IsoData".
 Select the image file as the input file and hit "OK". A box will appear that lists
parameters with the unsupervised classification by choosing different numbers of
classes and has a blank space after "Enter Output Filename". Name your
output file (something logical). Hit "OK".
 Select the output file and choose "Gray Scale". Choose "New image" to keep
the original color composite in an open window. Click on "Load Band".
 The unsupervised classification will appear as a false color image.
 Link the display windows to examine differences among the classifications and
input dataset.

 Try the default parameters then try following ISODATA parameters.

No. of classes: 7
Maximum Iteration: 5
Max stdv from Mean: 4
Max. Distance Error:

Guide Questions:
1. After filling up the desired parameters (above), answer the questions below by filling
up the table accordingly.

Questions Answer (default) Answer*

7. Number of classes
8. How many pixels were unclassified?
*for given parameters

2. What will happen to the output image if “maximum class stdv” is increased? If
decreased?
_____________________________________________________________________
_____________________________________________________________________
_____________________________________________________________________
 29

3. Discuss what have ISODATA method of unsupervised classification.

_____________________________________________________________________
_____________________________________________________________________
_____________________________________________________________________

4. Now, classify the image using “K-Means”. Click "Classification" in the main ENVI
menu bar.
 Select "Unsupervised”. Choose "K-means".
 Select the resized image file as the input file and hit "OK" . A box will appear that
lists parameters with the unsupervised classification by choosing different
numbers of classes and has a blank space after "Enter Output Filename".
Name your output file (something logical). Hit "OK".
 Link the display windows to examine differences among the classifications. You
may want to link the k-means to the isodata.

Guide Questions:

1. Compare K-Means method/technique with the ISODATA.

_____________________________________________________________________
_____________________________________________________________________
_____________________________________________________________________

V. Results and Discussion

VI. List of References

14. ENVI Tutorial

15. ENVI Help
16. ESRI (Environmental Systems Research Institute, Inc.). What is image
classification? Retrieved from ESRI website
http://desktop.arcgis.com/en/arcmap/latest/extensions/spatial-analyst/image-
classification/what-is-image-classification-.htm
 30

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 7

Pixel-based Supervised Classification

Title

I. Introduction:

In pixel-based supervised technique, the analyst assigns or selects representative

samples for each land cover class in the digital image, these pixels are called training
sites. The image processing software then uses training areas to identify the land cover
classes in the entire image. Out of the samples, the software generates signature profile,
of each class then classifies the entire image. The common supervised classification
algorithms are maximum likelihood and minimum-distance classification.
In this exercise, each method of supervised classification will be performed.

II. Objectives

At the end of this exercise student will be able to:

7. Discuss each method of supervised classification

8. Perform supervised image classification

III. Materials/Data
Landsat 7 (L7_IN_2005)

IV. Procedure/Preparation:

5. Perform image classification, subset image such that the upper-left and lower-right
(pixel) coordinates to be 4200, 3600 (samples) and 6500, 6500 (lines), respectively.
6. Original image may be closed. Perform supervised classification of the resized image.
7. Before performing supervised classification, it is required first to make training sites of
each class.
8. In the main image window, select Overlay Region of interest (ROI). The ROI tool will
pop up.
a. To define a region of interest, or training area, find an area on the image that you feel
represents a distinct spectral class. For example, you may choose a spot in the river
that is a specific color or you may choose a field.
 31

b. Use the mouse to define on the image a polygon that incorporates characteristic pixels
of the spot you chose.
c. Click on the left button of the mouse to mark the endpoints of sides of the polygon. To
close the polygon, click on the right button. Finish the procedure by double-clicking on
the right mouse button. The polygon will be filled in with the color red. In the ROI tool
box, the number of pixels incorporated in the red region will be listed.
d. If you want to add pixels from another area of the image into the ROI, simply draw
another polygon. The number of pixels in each ROI is listed.
e. Click “Edit” in the ROI Tool box and then Edit ROI Parameters dialog box appears.
Define name of this ROI in field of Name. Click “OK”

 Select additional regions of interest. Click "New Region" in the ROI Tool box and
repeat the procedure outlined above. This time the color will be green. The next
time, the color will be blue.
 Make up to a minimum of four classes (e.g. river, sea, bareland, vegetation). For
high accuracy, it is advised to make many training areas in each class, for example
for waterbody (landcover), training sites can be waterbody1, waterbody2, and so
on and so forth, or vegetation1, vegtation2, and so forth). Merging of these classes
will be performed in the next exercise.
 Save your ROIs to a file.
9. Once the training sites are done, checking of basic statistics associated per class is the
next step to do in order to check the variability of pixel values in a class.
10. In the main window, click Overlay Region of interest. The "ROI Tool" dialog box
will appear. (If it is already open, you don't need to do this step).
 Choose "File  Export ROIs to n-D Visualizer and select the image to be
classified. The "n-D Visualizer Input ROIs" dialog box will appear.
 Select all the classes and click "OK". Two displays will open, 1) the "n-D Visualizer"
window and 2) the "n-D Controls" dialog box.
 32

 The first will be black and the second will have 6 numbers, each of which
corresponds to a ETM+ band from the "pathum_etm" image. Move the white-cross
hair in the "n-D Controls" window with the cursor and click on band 2. It will be
highlighted. Click on band 3. A 2-D scatterplot will appear in the "n-D Visualizer"
that shows the pixel values in bands 2 and 3 for your classes.

 Experiment with other 2D scatterplots.

 Choose more than 2 bands. The control buttons at the bottom of the "n-D Controls"
window now appear as black. Click "Start". The pixels in the "n-D Visualizer" will
begin to rotate. The speed of the rotation can be controlled. To see where the axes
of the plot are that correspond to the different bands, go to "Options  Axes: On".
 Select all 6 bands in the "n-D Controls" dialog box to examine all 6 bands at once.

From n-D Visualizer and compare the result of classification some area is still overlapped.
Add two more classes to avoid or minimize overlapped. But if your old ROI include some
Grass and Wet Area, you have to delete that area first. Then make the supervised
classification with new ROI set.

Guide Questions:
1. Examine the statistics associated with the classifications.

Class No. Class Name Min. Max. Stdv. Mean Eigenvalues

 33

11. Once the training sites are done, classification is the next step to do.

12. Go to "Classification" in the main ENVI menu bar. Select "Supervised".

 Select “Maximum likelihood”. Choose the image file to be classified as the input
file and hit "OK". A box will appear that specifies the parameters for the
classification method you chose. The box will ask you to select the classes from
regions. You will use the ROIs that you created. Choose "Select All Items".
 The default for the probability threshold is that nothing is specified. All the pixels,
therefore, are classified in the default image. Specify a probability threshold of
0.90.
 Enter an output class filename. Make it logical. Also make sure that "No" is filled
in for "Output Rule Images".
 Select the output file for the classification image to display as Gray Scale when the
available bands list dialogue box appears. Make sure to load this as a new display
if you do not want to overwrite an existing window. You can link this display to any
of the other open displays for comparison.

Guide Questions:
1. Identify/compute the area per class in square meters.

Class No. Class Name Area (sq.m.)

V. Results and Discussion

VI. List of References

17. ENVI Tutorial

18. ENVI Help
19. ESRI (Environmental Systems Research Institute, Inc.). What is image
classification? Retrieved from ESRI website
http://desktop.arcgis.com/en/arcmap/latest/extensions/spatial-analyst/image-
classification/what-is-image-classification-.htm
 34

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 8

Post Classification
Title

I. Introduction:

Post classification tool in ENVI allows the analyst to classify rule images, to calculate
class statistics and confusion matrices, to apply majority or minority analysis to
classification images, to clump, sieve, and combine classes, to overlay classes on an
image, to calculate buffer zone images, to calculate segmentation images, and to output
classes to vector layers.

II. Objectives

At the end of this exercise student will be able to:

1. Apply majority or minority analysis to classification images

2. Clump, sieve, and combine classes in order to improve accuracy of classified
image

III. Materials/Data
Subset_image = subset image of Landsat 7 ETM, acquired 7 December 2001
Classification_map = classified image of the subset_image

IV. Procedure/Preparation:
1. Run ENVI. Then, open “classified_map” and “subset_image”
2. In main Menu, click Classification  Post Classification  Class Statistics.
1.1 Choose the classified image and click OK. Select the image used to produce
the classification and click OK.
1.2 Use the “Class Selection” dialog box to choose the classes for statistics. Click
on “ Select All Items” and then click OK.
1.3 Choose the statistics to be calculated (make sure to pick the covariance and
correlation matrices) in the Compute Statistics Parameters dialog box and
click OK. Several plots and reports will appear on the screen, depending on the
statistics options chosen.
 35

Guide Questions:
2. Identify/compute the area per class in square meters.

Class Name Area (sq.m.) % Total Area

Grassland
Urban area
Waterbody2
Waterbody3
Paddy field1
Paddy field2
Paddy field prep
Bareland

3. Combining sub-classes
Combine waterbody3 with waterbody2 and waterbody1, paddy field 1 with paddy
field 1 and paddy field 2 and paddy field prep.
3.1 To combine classes, click Classification  Post Classification  Combine
Classes. Choose one class to combine classes. Click OK.
3.2 Combine Class Parameters window appears then select an input class and
output class. Click Add combination, then combine the remaining sub-classes,
then click OK once all sub-classes are combined.
3.3 Combine classes output window appears. Select “Yes” for “Remove Empty
Classes?”, and select the output filename. The combine classification will then
process, and will display in Available band list dialog box.
3.4 Load the combined classification image. Select Tools -> Color Mapping ->
Class Color Mapping. Then change the combined class to the “new class
name” (something logical) in the class name field. Go to Options -> Save
changes.

Questions:
1. Update/compute the new area of the new class in square meters.

Class Name Area (sq.m.) % Total Area

Grassland
Urban area
Waterbody3
Paddy field1
Bareland

4. Majority/Minority Analysis. Majority/Minority Analysis tool is use to apply majority

or minority analysis to a classification image. Use majority analysis to change
spurious pixels within a large single class to that class. You enter a kernel size and
 36

the center pixel in the kernel will be replaced with the class value that the majority
of the pixels in the kernel has. If you select Minority analysis, then the center pixel
in the kernel will be replaced with the class value that the minority of the pixels in
the kernel has.

4.1 Go to Classification -> Post Classification -> Majority/Minority Analysis.

4.2 Choose on your “new” classified image (the all combined sub-classes, result of
# 3.4) when the Classification Input File dialog appears and click OK. The
Majority/minority parameters dialog box will appear with all the available class
listed under txt label “Select Classes”. Select all the classes and change the
kerner size to 3 x 3 kernel size then click OK.

Guide Questions:
1. Update/compute the new area of the new class in square meters.

Class Name Area (sq.m.) % Total Area

Grassland
Urban area
Waterbody3
Paddy field1
Bareland

5. Sieving of Classes. Isolated pixels may occur in classification images. This

problem can be solved by smoothing the image with a low-pass filter, but the class
information would be contaminated by adjacent class values. The problem
therefore, is addressed by using Sieve Classes, which removes isolated pixels.

5.1 Go to Classification  Post Classification  Sieve Classes. Select the

output from the Majority/minority (from # 4.2) operation when the Classification
Input File dialog appears and click OK. The Sieve Parameters dialog box will
appear with all of the available classes listed under the text label “Select
Classes”. Select all classes and click OK.

6. Clumping of Classes. Classified image also sometimes suffer a lack of spatial

coherency. This is manifested as speckle or holes in a classified area/image. This
problem is solved by using low-pass filter, but again the class information would be
contaminated by adjacent class values. The problem, therefore, is solved by using
“Clump Classes”, which combines adjacent similarly classified areas. Clumping is
typically done after sieving.

6.1 Go to Classification  Post Classification  Clump Classes. Select the

output from the sieve classes (from # 5.1) operation when the Classification
Input File dialog appears and click OK. The Sieve Parameters dialog box will
 37

appear with all of the available classes listed under the text label “Select
Classes”. Select all classes and click OK.

Guide Questions:
1. Update/compute the new area of the new class in square meters.

Class Name Area (sq.m.) % Total Area

Grassland
Urban area
Waterbody3
Paddy field1
Bareland

7. Exporting raster classified image to vector layer. Classified image can be

exported as vector layer and be open in any GIS software.
7.1 Select Classification  Post Classification  Classification to vector.
7.2 When the raster to vector input band dialog appears, select the desired input
classification file.
7.3 When the raster to vector parameters appears, select the desired class names
to convert to vector polygons by clicking on the class names.
7.4 Use the arrow toggle button to select output all of the classes to a single layer.
7.5 Select output to File. Click OK to make polygon vector layer for each class
selected.
7.6 New created vector will be added to Available vector list dialog. Click “Load
Selected” button.
7.7 In vector window parameter dialog, click Apply to load the created vector layer.
7.8 To export vector layer to Arcview layer, go to File  Export Layer to Arcview
Layer, then, enter the specific filename of the vector file.

V. Results and Discussion

VI. List of References

20. ENVI Tutorial

21. ENVI Help
22. ESRI (Environmental Systems Research Institute, Inc.). What is image
classification? Retrieved from ESRI website
http://desktop.arcgis.com/en/arcmap/latest/extensions/spatial-analyst/image-
classification/what-is-image-classification-.htm
 38

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 9

Accuracy Assessment
Title

I. Introduction:

Accuracy assessment is performed to check the accuracy of the classified image or

to compare it with the ground truth information. This process tells the quality of the
information derived from remotely sensed data. In ENVI, this tool is called Confusion
Matrix in which it compute overall accuracy, producer and user accuracies, kappa
coefficient, confusion matrix, and errors of commission and omission.
During the process, reference data or ground truth data is assumed to be 100%
correct. Referenced data are ROI that were gathered during field/ground validation or
extracted from the updated aerial photographs, or maps.
In this exercise, ROI from ground validation is already prepared for the purpose of
this exercise.

II. Objectives

At the end of this exercise student will be able to:

1. Determine the overall accuracy of the classified image

2. Solve overall accuracy, kappa coefficient, errors of commission and omission from
confusion matrix report

III. Materials/Data
Classification_map = classified image of the subset_image
Referenced_roi = ground truth data

IV. Procedure/Preparation:

1. From the Toolbox, select Classification  Post Classification  Confusion Matrix

Using Ground Truth ROIs. The Classification Input File dialog appears.
2. Select a classification input file and perform (optional) spatial and
spectral subsetting, then click OK. The Ground Truth Input File dialog appears.
The Match Classes Parameters dialog appears.
 39

3. Match the ground truth ROIs with the classification result classes by selecting the
matching names in the two lists and clicking Add Combination. The class
combinations are shown in a list at the bottom of the dialog. If the ground truth and
classification classes have the same names, they are automatically matched.
4. To remove a class match from the list, select the combination name. The two class
names reappear in the lists at the top of the dialog.
5. Click OK. The Confusion Matrix Parameters dialog appears.
6. Select the Pixels and/or the Percent check boxes.
7. Click the Yes or No toggle for Report Accuracy Assessment, and click OK.
8. The report shows the overall accuracy, kappa coefficient, confusion matrix, errors
of commission (percentage of extra pixels in class), errors of omission (percentage
of pixels left out of class), producer accuracy, and user accuracy for each class.
Producer accuracy is the probability that a pixel in the classification image is put
into class x given the ground truth class is x. User Accuracy is the probability that
the ground truth class is x given a pixel is put into class x in the classification image.

Guide Questions:
3. Fill-up the confusion matrix below

Urban Crop Range Water Total PA EO

Urban
Crop
Range
Water
Total
CA
EC

4. What is the overall accuracy? Kappa coefficient?

_________________________________________________________________

V. Results and Discussion

VI. List of References

23. ENVI Tutorial

24. ENVI Help
25. https://www.harrisgeospatial.com/docs/CalculatingConfusionMatrices.html
 40

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 10

Map Analysis
Title

I. Introduction:

Analysis can be performed in ENVI using its Band Math tool. This is done after image
processing in order to come up with analysis with the data for example, suitability analysis
and change detection analysis. In this exercise, these two analysis will be performed to
demonstrate the capabilities of Band Math tool in doing analysis.

II. Objectives

At the end of this exercise student will be able to:

1. Perform suitability analysis using Band Math tool

2. Perform change detection analysis using Band Math tool

III. Materials/Data
2011_map = land cover map for 2011
2013_map = land cover map for 2013

IV. Procedure/Preparation:

1. Given the land cover maps of a certain municipality for period 2011 and 2013. It
is required to perform change detection analysis using band math. Pixel values of
both images indicates as follows: 1 = water, 2 = forest, 3 = bareland, and 4 =
vegetation

2. In band math dialogue box, write the script:

b1 * 10 + b2
Pair “b1” with 2011 and “b2” with 2013, then execute. Examine then the statistics of
the new image.

Guide Questions:
5. What are the possible pixel values? What is the meaning of this pixel values?
 41

_________________________________________________________________
_________________________________________________________________
_________________________________________________________________
_________________________________________________________________

6. How much area change and not change? Fill up the table below.

Land cover change Area in Hectares

Forest to forest
Forest to water
Forest to bareland
Forest to vegetation
Vegetation to forest
Vegetation to water
Vegetation to bareland
Vegetation to vegetation
bareland to forest
bareland to water
bareland to bareland
bareland to vegetation
water to forest
water to water
water to bareland
water to vegetation

V. Results and Discussion

VI. List of References

26. ENVI Tutorial

27. ENVI Help
42
 43

Visayas State University

College of Engineering
Department of Geodetic Engineering
Visca, Baybay City, 6521-A, Leyte, Philippines

GEng 153 – Photo Interpretation & Remote Sensing

Laboratory Exercise No. 11

Decision Tree
Title

I. Introduction:

A decision tree is a type of multistage classifier that can be applied to a single image
or a stack of images. It is made up of a series of binary decisions that are used to
determine the correct category for each pixel. The decisions can be based on any
available characteristic of the dataset. For example, you may have an elevation image
and two different multispectral images collected at different times, and any of those images
can contribute to decisions within the same tree. No single decision in the tree performs
the complete segmentation of the image into classes. Instead, each decision divides the
data into one of two possible classes or groups of classes.
ENVI Classic provides a decision tree tool designed to implement decision rules, such
as the rules derived by any number of excellent statistical software packages that provide
powerful and flexible decision tree generators. Two examples that are used commonly in
the remote sensing community include CART by Salford Systems and S-PLUS by
Insightful. The logic contained in the decision rules derived by these software packages
can be used to build a decision tree classifier with ENVI Classic’s interactive decision tree
tool.

II. Objectives

At the end of this exercise student will be able to:

1. Discuss the capabilities of ENVI for image analysis

2. Classify image in decision tree

III. Materials/Data
IN_DEM = DEM file
L7_IN_2005 = landsat 7 image

IV. Procedure/Preparation:
 44

1. Load the landsat and DEM file in available band list.

2. Entering Decision Tree Rules. From the ENVI Classic main menu bar, select
Classification  Decision Tree  Build New Decision Tree. The ENVI Classic
Decision Tree dialog appears. By default, the decision tree tool starts with one
empty decision node that will divide the pixels in the dataset into two groups, using
whatever binary decision expression is entered into that empty node. The first
decision will be based on the Landsat image. To define the decision, click on the
decision node labeled Node 1. The Edit Decision Properties dialog appears.
3. In the Name field, type NDVI > 0.3. This is the text that will appear on the decision
node in the graphical view of the decision tree.
4. In the Expression field, type {ndvi} gt 0.3, then click OK. The Variable/File Pairings
dialog appears. The expression entered into the field tells ENVI Classic to divide
the pixels into those that have green vegetation and those that do not. NDVI is a
commonly used index of green vegetation abundance calculated from the red and
near-infrared bands of a multispectral image. The decision tree will calculate the
NDVI vegetation index for each pixel, and find all of the pixels that have values
higher than 0.3. The pixels with NDVI above 0.3 should have some green
vegetation.
5.
6. Pairing the Expression Variable with a File. In the Variables/Files Pairing dialog,
click once on the {ndvi} variable. The Select File to Associate with Variable dialog
appears.
7. Select the L7_IN_2005 image and click OK. This step tells the decision tree that
when evaluating this decision expression, the NDVI values should be calculated
from the L7_IN_2005 image. Because the wavelengths are known, ENVI Classic
will figure out which bands are needed in the NDVI calculation. If an image without
wavelengths in the header had been chosen, then you would have been prompted
to identify the red and near infrared bands for the NDVI calculation
8.
9. Entering Additional Rules. This is now a very simple decision tree classifier. Pixels
with NDVI greater than 0.3 will be included in Class 1 (the white class), and pixels
with NDVI less than or equal to 0.3 will be included in Class 0 (the black class).
You can enter additional decision rules to develop a more complicated
classification
10. Right-click on the Class 1 node and select Add Children to subdivide the pixels
with high NDVI into subcategories. Two new classes are automatically created
beneath Class 1.
11. The node previously labeled Class 1 is now blank. Click once on this node. The
Edit Decision Properties dialog appears.
12. In the Name field, type Slope < 20. In the Expression field, type: {slope} lt 20, then
click OK. This decision will divide the high NDVI pixels based on the steepness of
the slope. Right-click on Class 2 (the green terminal node) and select Add Children
to subdivide the pixels with high NDVI and low slopes into those with north facing
aspects and those without strongly north facing aspects. The node previously
 45

labeled Class 2 is now blank. Click once on this node. The Edit Decision Properties
dialog appears. In the Name field, type North. In the Expression field, type:
{aspect} lt 20 and {aspect} gt 340, then click OK.
13. Right-click on Class 0 (the black terminal node) and select Add Children to
subdivide the pixels with low NDVI values into those with values less than 20 in
band 4 (these will mainly correspond to water).
14. The node previously labeled Class 0 is now blank. Click once on this node. The
Edit Decision Properties dialog appears.
15. In the Name field, type Low B4. In the Expression field, type: b4 lt 20, then click
OK.
16. Executing the Decision Tree. The tree is now finished, but before it can be
executed, all of the variables used in the tree expressions must be paired with
image files.
17. In the Variables/Files Pairing dialog, click once on the {b4} variable. The Select
File to Associate with Variable dialog appears.

18. Select Band 4 from the L7_IN_2005 image and click OK.
19. Pair the following variables and files or bands as described in steps 1 and 2 above:
{slope} = IN_DEM
{aspect} = IN_DEM
20. From the ENVI Classic Decision Tree dialog menu bar, select Options  Execute,
or right-click in the blank background area of the ENVI Classic Decision Tree dialog
and choose Execute. The Decision Tree Execution Parameters dialog appears.
21. Select the L7_IN_2005 image to use as the base. The map projection, pixel size
and extent of the other image will be adjusted to match.
22. Enter an output filename for the classified image, and click OK.
 46

Guide Questions:
7. How many classes in all? Explain the meaning of each class.

Class No. Description

V. Results and Discussion

VI. List of References

28. ENVI Classic Tutorial on Decision Tree

29. ENVI Help