Expert

Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

Expert

O N - L I N E M A N U A L
Copyright  1982 - 1999 by ERDAS, Inc. All rights reserved.

Printed in the United States of America.

ERDAS Proprietary - Delivered under license agreement.


Copying and disclosure prohibited without express written permission from ERDAS, Inc.

ERDAS, Inc.
2801 Buford Highway, N.E.
Atlanta, Georgia 30329-2137 USA
Phone: 404/248-9000
Fax: 404/248-9400
User Support: 404/248-9777

Warning
All information in this document, as well as the software to which it pertains, is proprietary material of ERDAS, Inc., and is
subject to an ERDAS license and non-disclosure agreement. Neither the software nor the documentation may be reproduced in
any manner without the prior written permission of ERDAS, Inc.

Specifications are subject to change without notice.

Trademarks
ERDAS is a trade name of ERDAS, Inc. ERDAS and ERDAS IMAGINE are registered trademarks of ERDAS, Inc. Model
Maker, CellArray, ERDAS Field Guide, and ERDAS Tour Guides are trademarks of ERDAS, Inc. Other brands and product
names are trademarks of their respective owners.
Expert On-Line Manual

Knowledge Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Set Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Set Cell Size (Map) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Set Cell Size (Geo) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Knowledge Engineer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Hypothesis Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Rule Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Variable Editor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Classification Path Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Paste Query . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

iii
Knowledge Classification

Knowledge Classification
The Knowledge Classification application is the means by which knowledge bases constructed
with the Knowledge Engineer are processed. The Knowledge Classifier can be opened in Test
Mode to access accuracy of the knowledge base during construction or in its regular mode to
produce permanent output images.

The user interface allows you to input a limited set of parameters to control the use of the
knowledge base. The user interface is designed to lead you through panels of input parameters.
Output files which can be produced with Knowledge Classification in its standard mode include
the classification image, which can be a single value per class or a fuzzy set, a confidence image,
and a feedback image. The latter two images are optional in the non-test mode classification.
Temporary files for each of these images are made with Knowledge Classification in Test Mode.

Open the Knowledge Classifier in its regular mode by selecting Knowledge Classifier... on the
Classification menu.

Open Knowledge Classification in Test Mode by selecting Evaluate | Test Knowledge Base...

in the Knowledge Engineer menu or by clicking the Run Test Classification icon on the
Knowledge Engineer icon panel.

START BY SELECTING A KNOWLEDGE BASE FILE: This first panel of the


Knowledge Classifier is for entering the name of the existing knowledge base to be used for
classification.

Knowledge Base: Enter the name of the knowledge base here.

SELECT THE CLASSES OF INTEREST:


Available Classes: The classes shown in this window are all of the hypotheses in the
knowledge base which have been specified as output classes. Classes can be chosen
from this list to add to the list of selected classes on the right to be included in the output
classification.

1
Knowledge Classification

All enabled classes are placed in the Selected Classes list by default.

Add -> Click this button to add a highlighted class in the Available Classes list to the
Selected Classes list on the right. Click on the name of the class in the Available Classes
list to highlight it before clicking the Add button.

Add All Click to add all classes that appear in the Available Classes List to the Selected
Classes list.

Remove <- Click to remove a highlighted class in the Selected Classes list. Click on the
name of the class in the Selected Classes list to highlight it before clicking the Remove
button.

The class will remain in the Available Classes list.

Remove All Click to remove all classes from the Selected Classes list. The classes will
remain in the Available Classes list.

Selected Classes: All classes in this list will be processed by Knowledge Classifier. Use
the buttons between the Available Classes and Selected Classes lists to add or remove
classes for classification.

SELECT THE INPUT DATA FOR CLASSIFICATION: This panel is displayed when
one or more variables have been set to Leave Undefined (Prompt Analyst). There is a
checkbox in the Knowledge Engineer Variable Properties for this purpose.

If enough entries appear in this panel to more than fill the entry area, the entries become a
scrolling list.

Type: The Variable type is listed below this heading.

Variable: The Variable Name appears here.

Value: Below here is the file name entry for the variable as well as the Layer selection
and the Attribute of the variable to be used. These two parameters can be selected from
the options on the popup list for each.

Info: This text entry field contains any additional information provided by the Knowledge
Base producer to describe what file is to be entered as the current input.

SELECT CLASSIFICATION OUTPUT OPTIONS:

2
Knowledge Classification

Best Classes Per Pixel Set the number of classes per pixel in the Output Classified
Image. Choosing one class per pixel will produce a standard thematic image. Choosing
more than one class will produce a fuzzy set of classes in a multi-band image. The best
class per pixel will be in the first layer of the output classification image, the second best
class in layer two, and so on. In the case of two or more classes having the same
confidence, they are placed in layers in the same order they appear in the decision tree.

Output Classified Image: Enter a name for your output classification image here. The
number of bands in this output image is defined by entry for the number of Best Classes
Per Pixel.

This image is automatically produced as a temporary file in Knowledge Classification Test


Mode. Filename entry is inactive in Test Mode.

Produce Confidence Image Check this box if an Output Confidence Image is desired.

The default for this checkbox can be set to on or off in Preferences for Knowledge
Classifier.

This image is automatically produced as a temporary file in Knowledge Classification Test


Mode. The checkbox and filename entry are inactive in Test Mode.

Output Confidence Image: Enter a name for the Confidence Image here. This text field
remains inactive if the Produce Confidence Image checkbox is unchecked.

Filename entry is inactive in Test Mode.

Produce Feedback Image Check this box to produce an Output Feedback Image.

The default for this checkbox can be set to on or off in Preferences for Knowledge
Classifier.

This image is automatically produced as a temporary file in Knowledge Classification Test


Mode. The checkbox and filename entry are inactive in Test Mode.

Output Feedback Image: Enter a name for the Feedback Image here. This text field
remains inactive if the Produce Feedback Image checkbox is unchecked.

Filename entry is inactive in Test Mode.

Output Map Projection Same As: When variables within a knowledge base have
different projections, a projection for the output image must be chosen here. The
projection is chosen by selecting one of the knowledge base images from the popup list.
The projection for that image is used for the output image.

Area: The method of determining the area for the output image(s) is selected here.

3
Knowledge Classification

Window Selecting Window from the popup list will allow one of the following options
to be chosen: Union, Intersection, or Specified (by coordinates). Clicking the Set...
button after selecting Window opens the Set Window dialog.

AOI Selecting AOI from the popup list and clicking the Set... button Window will open
the Choose AOI dialog.

Set... Click this button to open the dialog for the chosen Area type, either Window or AOI.

Cell Size: Select the method for determining the Cell size. If Specify is chosen, click the
activated Set... button to open the Set Cell Size dialog.

Minimum The smallest cell size found in all of the input images will be used for the
output image cell size.

Maximum The largest cell size found in all of the input images will be used for the
output image cell size.

Specify Select this option and click the Set Button... to open the Set Cell Size dialog
where a new cell size may be entered for X and Y.

Set... Click this button to open the Set Cell Size dialog. This button is inactive until the
Specify option is chosen in the popup list for Cell Size.

THE FOLLOWING FILES COULD NOT BE FOUND: This panel is displayed only
if files specified in the knowledge base are not found when checked by Knowledge Classifier.

Files listed in the knowledge base that cannot be found by Knowledge Classifier are listed
here. If the file is available but it is in a different location or has a different name, enter the file
name and/or new path to continue.

If multiple files are missing a scrolling list of them is shown here.

An appropriate file must be entered here to produce classes from hypotheses that use this
particular file in the knowledge base. An alternative, if the appropriate file is not available, is
to remove any associated classes from the output class list and continue the classification for
the remaining classes.

Type: The file Type of the missing file is shown here.

Variable: The Variable name of the missing file is shown here

Value: The file name for the missing file is entered here.

Next Click to continue to the next panel of Knowledge Classification.

Previous Click to return to the previous panel of Knowledge Classification.

4
Knowledge Classification

OK Click to accept the current parameters and begin Classification processing.

Batch Click to include this process in a batch file.

Cancel Close the Knowledge Classification Dialog without running a classification.

Help Click to open this On-line Help document.

Save As... Click this button to save the knowledge base with the current parameters.
For more information, refer to the ERDAS IMAGINE Expert Classifier User’s Guide.

5
Set Window

Set Window
This dialog allows you to set an output window and coordinates for the classification output layer.

This dialog is opened when you click the Set... button after selecting Specify from the Cell Size
popup list on the Select Classification Output Options panel of the Knowledge Classifier.

Set Window To: Set the output window to union or intersection of all layers, or set it
explicitly.

Union Set the output window to Union of all layers.

Intersection Set the output window to intersection of all layers.

Specify Set the output window explicitly.

RADIOBUTTON
Map Select to use map coordinates to specify the output window.

File Select to use file coordinates to specify the output window. The file coordinates used
are from the file selected in the Output Map Projection Same As popup list or the first listed
variable if no file has been selected in the popup list.

From Inquire Box Click to define a window for the data by using the Viewer Inquire Box.
When you click this button, the coordinates below are updated with the coordinates of the
cursor box in the viewer.

To change these coordinates, you can move and/or resize the cursor box in the Viewer, then
click this button again.

The image you are using and the inquire box must already be displayed in the Viewer in order
to use the From Inquire Box option. Otherwise, you may manually enter coordinates in the
fields below.

UL X: Select the upper left X coordinate of the output image.

UL Y: Select the upper left Y coordinate of the output image.

LR X: Select the lower right X coordinate of the output image.

LR Y: Select the lower right Y coordinate of the output image.

Cancel Click to close this dialog without making any changes.

6
Set Window

OK Click to set these parameters and close this dialog.

Help Click to open this On-Line Help document.

7
Set Cell Size (Map)

Set Cell Size (Map)


A new cell size and its units, meters or feet, may specified in this dialog.

This dialog is opened when you click the Set... button after selecting Specify from the Cell Size
popup list on the Select Classification Output Options page of the Knowledge Classifier. This
particular Set Cell Size dialog opens if all of the variable layers, or the layer selected in the Output
Map Projection Same As popup list, are georeferenced to a projection or coordinate system (not
Geographic (Lat/Lon)).

X: Select the X cell size of the output image.

Y: Select the Y cell size of the output image.

Units: Click this popup list to select output map units.


meters The cell size of the output image will be in meters.

feet The cell size of the output image will be in feet.

OK Click to set these parameters and close this dialog.

Cancel Click to close this dialog without making any changes.

Help Click to open this On-Line Help document.

8
Set Cell Size (Geo)

Set Cell Size (Geo)


A new cell size and its units, degrees or radians, may specified in this dialog.

This dialog is opened when you click the Set... button after selecting Window from the Area
popup list on the Select Classification Output Options page of the Knowledge Classifier. This
particular Set Cell Size dialog opens if all of the variable layers, or the layer selected in the Output
Map Projection Same As popup list, are georeferenced to Geographic (Lat/Lon).

X: Select the X cell size of the output image.

Y: Select the Y cell size of the output image.

Units: Click this popup list to select output map units.


degrees The cell size of the output image will be in degrees.
radians The cell size of the output image will be in radians.

OK Click to set these parameters and close this dialog.

Cancel Click to close this dialog without making any changes.

Help Click to open this On-Line Help document.

9
Knowledge Engineer

Knowledge Engineer
This is the main dialog of the Knowledge Engineer program. All of the main functions for creating
and testing a knowledge base are accessible from the dialog. The dialog consists of menu and
icon panels, a knowledge tree overview, hypotheses, rules and variables listings, and the main
knowledge base display/edit window.

This dialog is opened by selecting Knowledge Engineer... from the Classification menu.

File
New Create a new knowledge base. This opens a new, empty, Knowledge Engineer
window while leaving the current Knowledge Engineer window open.

Open... Select to open a knowledge base for editing. The Open Classification Knowledge
Base dialog appears.

When a different knowledge base is chosen, it replaces the current knowledge base in the
same Knowledge Engineer dialog.

Close Close the current Knowledge Base.

Save Save the current Knowledge Base. This option is inactive (“grayed out”) until the
knowledge base has first been saved using the Save As... option.

Save As... Select this to save a knowledge base which has not yet been saved or to save
it to a new name. The knowledge base can be saved as the default Classification
Knowledge Base with its .ckb extension for future use with Knowledge Engineer or
Knowledge Classifier.

The knowledge base can also be saved to Encapsulated PostScript (.eps) or PostScript
(.ps) to be printed using PostScript printers.

Revert to Saved Revert knowledge base to the last saved version.

☞ Any edits which have been made since the last save are lost!
Print... Print the current knowledge base. Printing the knowledge base requires a
PostScript printer. A PostScript 2.0 (.ps) file of the knowledge base is sent to the printer.

To print an Encapsulated PostScript version of the knowledge base to a PostScript 3.0


printer, save the knowledge base, using File | Save As..., to an .eps file. Then print the
resulting Encapsulated PostScript file outside of ERDAS IMAGINE.

Close All Close all Knowledge Engineer windows.

Edit

10
Knowledge Engineer

Undo Undo your last edit.

Cut Cut the selected hypotheses and/or rules and conditions from the Knowledge Base.

Copy Copy the selected hypotheses and/or rules and conditions to the paste buffer.

Paste Copy the paste buffer into the Knowledge Base.

New Hypothesis... This opens Hypothesis Properties so that a new hypothesis can be
defined. After clicking Apply in Hypothesis Properties, the new hypothesis will appear
below the last item in the knowledge tree and in the hypotheses list.

New Rule... Selecting this will open Rule Properties so that a new rule can be defined.
After clicking Apply in Rule Properties, the new rule and any added conditions will appear
below the last item in the knowledge tree. The rule will appear in the Rules list and any
variables added to the Rule will appear in the Variables list.

New Variable... Add a new variable. Selecting this will open Variable Editor so that the
variable can be defined. After clicking Apply in Variable Editor, the new variable will be
added to the variable list.

The new variable can then be easily added to a rule by selecting it from the popup list
which appears when clicking a cell in the Variable column of the Rule Properties dialog.

Enable All Enable all Nodes in the Knowledge Base. Re-enabled nodes appear again
in “full color” in the edit window. All enabled nodes are used by the Knowledge Classifier.

Disable All Disable all Nodes in the Knowledge Base. Disabled nodes appear to be
“grayed out” in the edit window. Disabled nodes are ignored by the Knowledge Classifier.

Disabling nodes is a good way to test the effect of removing nodes from a classification
without actually deleting them.

Delete All Disabled Delete all disabled Nodes in the Knowledge Base.

Be careful. All disabled nodes will be deleted. Some disabled nodes may not be visible in
the edit window without panning.

Clear All Work Files... Clear all temporary work files for models and programs.

Clear work files if you wish to allow new temporary files to be generated. Otherwise, the
semi-permanent existing work files will be used, if available for the knowledge base
variables, at the time a variable’s model or program is run. To clear a work file for an
individual variable, click the Clear Work File button in Variable Editor for that particular
variable.

Evaluate

11
Knowledge Engineer

Test Knowledge Base... Start the Knowledge Classifier with the Current Knowledge
Base. Knowledge Classifier will open in Test Mode.

The Output Classified Image, Confidence Image, and Feedback Image will all be
temporary files when produced in Test Mode.

Classification Pathway Cursor... Select this to open the Classification Path Information
dialog. The Pathway Cursor for the Test Classification opens in the Viewer containing the
test classification image.

When the pathway cursor is placed on a non-zero pixel in the test classification image, the
section in the knowledge tree by which this pixel was classified is highlighted in the
Knowledge Engineer edit and overview windows.

No edits can be made in the knowledge base while the Classification Pathway Cursor is
open.

Help
About Knowledge Engineer... Click to open this On-Line Help document.

Open a Knowledge Base for Editing. Selecting this will open the Open Classification
Knowledge Base dialog. When another knowledge base is chosen, it will replace the current
knowledge base in the same Knowledge Engineer dialog.

Save the current Knowledge Base. This option is inactive (“grayed out”) until the
knowledge base has first been saved using the Save As... option.

Print the current Knowledge Base. Printing the knowledge base requires a PostScript
printer. A PostScript 2.0 (.ps) file of the knowledge base is sent to the printer.

To print an Encapsulated PostScript version of the knowledge base to a PostScript 3.0 printer,
save the knowledge base, using File | Save As..., to an .eps file. Then print the resulting
Encapsulated PostScript file outside of Imagine.

Start the Knowledge Classifier in Test Mode with the Current Knowledge Base.

The Output Classified Image, Confidence Image, and Feedback Image are all temporary files
when produced in Test Mode.

12
Knowledge Engineer

Select this icon to open the Classification Path Information dialog. The Pathway Cursor
for the Test Classification opens in the Viewer containing the test classification image.

When the pathway cursor is placed on a non-zero pixel in the test classification image, the
section in the knowledge tree by which this pixel was classified is highlighted in the
Knowledge Engineer edit and overview windows.

No edits can be made in the knowledge base while the Classification Pathway Cursor is open.

Undo your last edit.

Cut the selected Items from the Knowledge Base. Select items in the edit window using
the pointer tool, then click this icon to delete them.

Copy the selected items to the paste buffer. Select items in the edit window using the
pointer tool, then click this icon to copy them.

Copy the paste buffer into the Knowledge Base.

Use this tool to select nodes in the Knowledge Base edit window. Click in a hypothesis
or rule graphic to select that node and all items beyond it (conditions for the node) for that
section of the knowledge tree.

Click this icon, then click within the Knowledge Engineer edit window to place a new
hypothesis graphic at the selected location. The new hypothesis can be placed in a blank area
of the edit window, on an existing rule, or on an existing hypothesis.

Placing the new hypothesis on an existing hypothesis replaces the existing one (An Attention
window asks if you really want to overwrite the existing hypothesis).

Click this icon, then click within the Knowledge Engineer edit window to place a new
rule graphic at the selected location. The new rule can be placed in a blank area of the edit
window, on an existing hypothesis, or on an existing rule.

Placing the new rule on an existing rule replaces the existing one (An Attention window asks
if you really wish to overwrite the existing rule).

13
Knowledge Engineer

Use the Enable Node graphic tool to enable sections of the knowledge base which have
been previously disabled. Click on the disabled node in the edit window to enable it.

➲ All enabled nodes are used by Knowledge Classifier during classification.

Use the Disable Node graphic tool to disable sections of the knowledge base. Click on
a node in the edit window to disable it.

Disabling nodes is a good way to test the effect of removing nodes from a classification
without actually deleting them.

Indicates that the current tool is locked and remains active until you select another tool.

Click to lock the current tool.

Hypotheses All hypotheses are listed in this folder.

Rules All rules are listed in this folder.

Variables All variables are listed in this folder.

Add a new item to the list in the currently selected folder.

Show properties of the selected item on the list. The appropriate Properties dialog
opens.

Delete the selected item from the list.

Click this icon to highlight the next occurrence of a selected hypothesis or rule in the
main display/edit window and in the overview window. If the current occurrence is the only
one, the current item remains highlighted.

For more information, refer to the ERDAS IMAGINE Expert Classifier User’s Guide.

14
Hypothesis Properties

Hypothesis Properties
The Hypothesis Properties dialog allows you to name a hypothesis, indicate whether it should be
a final output class, and specify a color if it is an output class.

This dialog may be opened by double-clicking on a hypothesis graphic, selecting a hypothesis in


the hypothesis folder and clicking the Show Properties of Item on List icon, double-clicking a
hypothesis or from the Knowledge Engineer Edit menu.

Name: Enter a name for the hypothesis here. This name appears in knowledge base
graphics and it is the title of the output class if applicable.

Create an Output Class Check this checkbox if the hypothesis is an output class in the
thematic image resulting from Knowledge Classifier.

Color: This section is enabled when the checkbox for Create an Output Class is checked.
Grayscale When this default option is selected, the hypothesis color is in grayscale. The
gray level assigned is determined in the end by the number of classes which have been
chosen to be grayscale. As output class hypotheses are added that are set to grayscale,
the gray assigned to each class is evenly divided on a scale of 255 - 0, starting at white
and ending at black.

➲ This scaling occurs whether or not other classes are assigned specific colors.
Specify Select this radio button to assign a color to an output class hypothesis.

Colorpatch This color patch shows the current color for the output class. Click the
Chooser button to select a different color for the class.

Apply Click to apply the current specifications to the hypothesis.


Close Click to close the dialog. Closing without applying changes leaves the hypothesis
in its previous state.

Help Click to display this On-Line Help document.

15
Rule Properties

Rule Properties
The Rule Properties dialog provides a way to name a rule, enter confidence levels, and enter
variables and values with their relationships for the rule.

This dialog may be opened by double-clicking a rule graphic, selecting a rule in the rule folder
and clicking the Show Properties of Item on List icon, or double-clicking a rule in the rule folder.

Name: Enter a name for the rule here. This name appears in Knowledge Engineer
graphics.

Rule Confidence: A confidence level for the rule is entered in this section. The
confidence level scale ranges from 0 - 1.

Compute from Conditions Selecting this radio button (which is selected by default)
allows the program to calculate the confidence based on the confidence of each condition
associated with this rule.

Specify Selecting this radio button allows a specified number to be used for the
confidence level.

( number field ) Type or use the nudgers to enter a specific confidence level value here.
The 0 - 1 range goes from lowest to highest confidence.

List of Conditions: All conditions immediately connected to a rule are listed in the
CellArray.

( CellArray ) Existing Variables and their Relation, Value, and Confidences are listed in
the CellArray. New variables, relations, and values can be entered here by right-clicking
within the appropriate cells to reveal pop-up menus.

The pop-up menu for Variable lists existing hypotheses and variables, which can be
selected as variables for the current rule, or New Variable... may be selected. Choosing
New Variable... opens the Variable Properties dialog.

The pop-up menu for Relation consists of As Is, ==, !=, >, >=, <, and <=. The entry relates
the variable to the value and the resulting equation indicates what makes this current
condition a true statement. If an existing hypothesis has been chosen as the variable, the
menu lists only As Is, ==, and !=.

If the variable is not an existing hypothesis, the pop-up menu for Value lists the existing
variables, which can be selected as values for the current rule, or Other... can be selected.
Selecting Other... allows the entry of numerical values. If an existing hypothesis has been
chosen as the variable, the options which appear for the right-click menu are As Is, True,
and False since conditions are set by the existing hypothesis.

16
Rule Properties

Apply Click to apply the current specifications to the rule.

Close Click to close the dialog. Closing without applying changes leaves the rule in its
previous state.

Help Click to display this On-Line Help document.

17
Variable Editor

Variable Editor
The Variable editor allows for the definition of variable objects to be used in rule conditions.

This dialog may be opened by selecting a variable in the Variable folder and clicking the Show
Properties of Item on List icon, double-clicking a variable in the Variable folder, double-clicking
a variable or empty variable cell in the Rules Properties CellArray, or from the Knowledge
Engineer edit menu.

Variable Name: Enter a name for the variable here. This name appears in the Knowledge
Engineer graphics.

Optional If this checkbox is checked, Knowledge Classifier uses the variable if it is


available in its specified location or ignores it and continue processing if the variable is not
present.

Variable Type:
Raster Select this option if the variable is to be one of the raster variable options (an
image or vector layer), or the raster result of a graphic model or program.

Scalar Choose this option if the variable is a scalar number or a scalar result of a graphic
model or program.

Data Type: Choose the appropriate Data Type for the variable considering the type of file
it is and its use in the classification.

Integer Make Integer the Data Type.

Float Select to make Float the Data Type.

Boolean Make Boolean the Data Type.

Raster Variable Options: These options are available if Raster is chosen as the
Variable Type above. A different interface appears below the radio buttons depending on
which of these options is chosen. Select one of the following:

Imagery Select the imagery option for a raster image. These images can be any
supported raster format and projection. A mixture of various raster formats and
projections is acceptable. The interface for these options is explained in Raster Variable
Options:Imagery.

Feature Select this option to enter an Arc Coverage, an SDE Vector Layer, or a
Shapefile. The interface for these options is explained in Raster Variable
Options:Feature.

18
Variable Editor

➲ Note that Feature layers can only be used in Knowledge Engineer as input to models.
Rasterization of any Feature layers occurs within the Modeler. The interface for these options
is explained in:

Graphic Model Choose this option to specify the output of a model as a variable.

The graphic model runs within Knowledge Classification to produce the variable raster file
“on-the-fly.” The interface for these options is explained in Raster Variable
Options:Graphic Model.

Program Select this option to specify the output of a program as a variable. The model
can be an ERDAS IMAGINE program that can run in a command line, non-interactive
mode. The program is initiated by Knowledge Classifier and becomes part of the
classification process. The interface for this option is explained in Raster Variable
Options:Program.

➲ The interface for these options is explained in Raster Variable Options:Program.

Scalar Variable Options: Scalar Variables can be defined by an explicit value or by


the output of a graphic model or program. The section of dialog below the radio buttons
change depending on what option is selected.

Value Select this radio button to enter an Explicit Value for the scalar. The interface for
these options is explained in Scalar Variable Options:Value.

Graphic Model Select this option to enter a Graphic Model with scalar output. The
interface for these options is explained in Scalar Variable Options:Graphic Model.

Program Choose this radio button to enter a Program with scalar output. The interface
for these options is explained in Scalar Variable Options:Program.

Apply Click to Apply the current parameters to the variable.


Close Click to Close this dialog.
☞ Changes that have not been applied are lost.

Help Click to open this On-line Help document.

19
Variable Editor

Raster Variable Options:Imagery

Leave Undefined (Prompt Analyst) This checkbox appears for the Imagery and
Feature Raster Variable Options. Selecting it allows you to have the end user running
Knowledge Classifier provide the image or feature for this variable. During test classifications,
you provide the file names for the undefined variables in order for the hypotheses using these
variables to work.

Info: This field becomes active when the Leave Undefined box is checked. It provides a
place for you to add directions for the person running Knowledge Classifier to determine what
file should be used for this variable.

Select Image: This field is present when Imagery has been chosen as the Raster
Variable Option. Enter the name of the image here.

Layer: Select a layer of the selected image for the variable.


If multiple layers are needed to describe a rule, they should be entered as separate variables
or entered as hard-coded parts of a model.

Use: Select either the Cell Value, or the Raster Attribute of the image, to define the variable.
The cell value or attribute chosen is used in conjunction with Relation and Value in Rule
Properties when the variable is made a part of a rule.

Raster Variable Options:Feature

Leave Undefined (Prompt Analyst) This checkbox appears for the Imagery and
Feature Raster Variable Options. Selecting it allows you to have the end user running
Knowledge Classifier provide the image or feature for this variable. During test classifications,
you have to provide the file names for the undefined variables in order for the hypotheses
using these variables to work.

Info: This field becomes active when the Leave Undefined box is checked. It provides a
place for you to add directions for the person running Knowledge Classifier to determine what
file should be used for this variable.

Select Feature: This field is present when Feature has been chosen as the Raster
Variable Option. Enter the name of the image here.

Raster Variable Options:Graphic Model

Graphic Model: Enter the name of the graphic model.

20
Variable Editor

The Knowledge Engineer requires that models be graphical model (.gmd) files. The model
may have multiple inputs but should have only one output.

Edit Model... Click to open the Modeler for the selected graphic model.

Output: The available output from the selected model is listed here.

Layer: Select the layer of the model output image to use for the variable.

Use: Cell Value of the model output image is used.

Define Model Input Nodes with Variables: This CellArray provides information on
the files that are used as inputs in the model. The Input Node name and type are provided as
well as the Variable Name for the input layer.

Clear Work File Click to clear the Work File for this model. New temporary files are
generated.

Raster Variable Options:Program

Program: Enter the name of the external program.


This allows for the integration of other ERDAS IMAGINE command line programs, such as
the classifiers. However, these external programs are not restricted to ERDAS IMAGINE
programs. The Expert Classifier can also run any other program provided it can be run in a
non-interactive mode, that it accepts command line arguments, and one of its command line
arguments is the output filename.

To create a program variable you must provide the program name, the number of arguments,
and which argument is the output filename. This information can be found in the IMAGINE
Command and Function Syntax reference. The CellArray must be populated with valid
arguments. Like the model variables, the program variable may use other variables in the
knowledge base as inputs to the program. The Var column in the CellArray indicates whether
the argument is a knowledge base variable or not. The program variable assumes the
variable name for referencing in the rest of the knowledge base.

Number of Arguments: Enter the number of arguments used by the external program.

Output Argument: Select the number of the argument in the external program which
serves as the output filename. This number must coincide with the row number of the
filename argument in the CellArray below.

21
Variable Editor

( CellArray ) In the CellArray, enter the arguments which appear in the external program
in the cells in the Arguments column. If a current knowledge base variable is used as an
argument, it is indicated in the Var column.

Clear Work File Click to clear the work file for this program. New temporary files will be
generated.

Scalar Variable Options:Value

Leave Undefined (Prompt Analyst) Check this box to have the end user provide the
scalar to be used for the variable.

Info: This field becomes active when the Leave Undefined checkbox is checked. It
provides a place for you to add directions for the person running Knowledge Classifier to
determine what file or value should be used for this variable.

Enter value: This entry appears if Integer or Float is chosen as the Data Type for the
scalar. Enter the explicit scalar value here.

Select value: This entry appears if Boolean is chosen as the Data Type for the scalar.
True Select to define the Boolean scalar entry as True.

False Select to define the Boolean scalar entry as False.

Scalar Variable Options:Graphic Model

Graphic Model: Enter the name of the graphic model.


The Knowledge Engineer requires that models be graphical model (.gmd) files. The model
may have multiple inputs but should have only one output.

Edit Model... Click to edit the model in Modeler.

Scalar Output: The available output for this model is listed here.

Define Model Input Nodes with Variables: This CellArray provides information on
the files that are used as inputs in the model. The Input Node name and type are provided as
well as the Variable Name for the input layer.

Clear Work File Click to clear the Work File for this model. New temporary files are
generated.

22
Variable Editor

Scalar Variable Options:Program

Program: Enter the name of the external program.


This allows for the integration of other ERDAS IMAGINE command line programs, such as
the classifiers. However, these external programs are not restricted to ERDAS IMAGINE
programs. The Expert Classifier can also run any other program provided it can be run in a
non-interactive mode, that it accepts command line arguments, and one of its command line
arguments is the output filename.

To create a program variable you must provide the program name, the number of arguments,
and which argument is the output filename. The CellArray must be populated with valid
arguments. Like the model variables, the program variable may use other variables in the
knowledge base as inputs to the program. The Var column in the CellArray indicates whether
the argument is a knowledge base variable or not. The program variable assumes the
variable name for referencing in the rest of the knowledge base.

Number of Arguments: Enter the number of arguments used by the external program.

Output Argument: Select the number of the argument in the external program which
serves as the output filename. This number must coincide with the row number of the
filename argument in the CellArray below.

( CellArray ) In the CellArray, enter the arguments which appear in the external program
in the cells in the Arguments column. If a current knowledge base variable is used as an
argument, it is indicated in the Var column.

Clear Work File Click to clear the Work File for this model. New temporary files are
generated.

23
Classification Path Information

Classification Path Information


This dialog presents information for an output layer of Knowledge Classifier. The dialog is
available for all test classifications conducted through the Knowledge Engineer. They are only
available for non-test mode classifications if an Output Feedback Mode Image has been created
(one of the options on the Select Classification Output Options panel of the Knowledge Engineer
Test Classification dialog).

Open the Classification Pathway Cursor by selecting Evaluate | Classification Pathway


Cursor... from the Knowledge Engineer menu or the Start Classification Pathway Feedback
Mode icon on the Knowledge Engineer icon panel.

Map Show the map coordinates for the test classification.

File Show the file coordinates for the test classification.

Lat/Lon Show Latitude/Longitude coordinates.

X: X coordinate value for the cursor position.

Y: Y coordinate value for the cursor position.

Meters/Feet, Pixels, Degrees Units of Coordinates are shown here. The units are
automatically updated to correspond to the choice of Map, File or Lat/Lon coordinates chosen
above.

The CellArray indicates the Class Name, Class Value, and Confidence Level for the current
pixel.

Move Cursor up by increment amount

Move Cursor down by increment amount

Move Cursor to Center of Window

Move Cursor left by increment amount

Move Cursor right by increment amount

24
Classification Path Information

( number field ) Set Increment Value

Map Increment Units

File Increment Units

( color patch ) Click this button to change the color of the path highlight in the edit window.
To change the color of the Classification Pathway cursor, select Utility | Selector Properties
in the viewer where the classification image is displayed.

Close Close Cursor Information Dialog

Help Cursor Information Help

25
Paste Query

Paste Query
The question of whether to place a Copy Of or Reference To a pasted hypothesis must be
answered before the hypothesis is added to the knowledge tree.

A copy of the hypothesis acts as a separate entity and can be edited without affecting the
hypothesis from which it was copied. If the new hypothesis is made a reference to the hypothesis
from which it was copied, any changes made to the reference or original hypothesis (or their
conditions) are automatically applied to both.

This dialog appears in response to adding a copy of an existing hypothesis to the knowledge
tree.

Paste a Copy Of or Reference To the Hypothesis?

Copy Click to make the hypothesis a copy of the original hypothesis and close the dialog.

Reference Click to make the hypothesis a graphical reference to the original hypothesis
and close the dialog.

Cancel Click to close this dialog without pasting a hypothesis copy or reference.

Help Click to open this On-Line Help document.

26

You might also like