This code example performs an end-to-end SVM classification using ENVI API objects. It performs the following steps:

• Extracts an ENVIExamples object from an attribute image and training data ROIs.
• Normalizes the examples to a common range of data values.
• Shuffles the examples to create a random distribution of data.
• Splits the examples into training and evaluation sets.
• Uses an iterative trainer to train a SVM supervised classifier.
• Plots a loss profile.
• Prints a confusion matrix and accuracy metrics from the trained classifier.
• Runs the classifier.
• Displays the classification image.

Follow these steps to run the code example:

1. Get the sample data files from our website or from the ENVI Resource DVD in the classification directory. Copy these files to your local drive.
2. Copy and paste the code into the IDL Editor.
3. Update the following lines to include the full path where you saved the sample data:
   • file = 'AttributeImage.dat'
   • ROIfile = 'TrainingDataROIs.xml'
4. Update the following line to include the full path where you will save the trained classifier:
   • classifierURI = 'TrainedSVMClassifier.epo'
5. Save the file as SVMClassificationUsingObjects.pro.
6. Compile and run the code.

PRO SVMClassificationUsingObjects

COMPILE_OPT IDL2

; Start the application

e = ENVI()

; Open an attribute raster to classify

file = 'AttributeImage.dat'

raster = e.OpenRaster(file)

; Open training data ROIs

ROIfile = 'TrainingDataROIs.xml'

rois = e.OpenROI(ROIfile)

; Extract examples and class values

outExamples = ENVIExtractExamplesFromRaster(raster, rois)

; Normalize the data

normalizedExamples = ENVIApplyGainOffsetToExamples( $

  outExamples, $

  OUTPUT_OFFSET=offset, $

  OUTPUT_GAIN=gain)

Print, 'Gain: ',gain

Print, 'Offset: ',offset

; Shuffle the examples to create a random distribution

shuffledExamples = ENVIShuffleExamples(normalizedExamples)

; Split the examples into training and evaluation sets

splitExamples = ENVISplitExamples(shuffledExamples, $

  SPLIT_FRACTION=0.8)

; Define the SVM classifier inputs

classifier = ENVISVMClassifier( $

  NATTRIBUTES=outExamples.NATTRIBUTES, $

  NCLASSES=outExamples.NCLASSES, $

  CLASS_NAMES=outExamples.CLASS_NAMES)

; Define the trainer inputs

trainer = ENVIIterativeTrainer( $

  CONVERGENCE_CRITERION=0.0001, $

  MAXIMUM_ITERATIONS=1)

; Train the classifier

ENVITrainClassifier, trainer, classifier, splitExamples[0], $

  LOSS_PROFILE=lossProfile

; Save the trained classifier to disk

classifierURI = 'TrainedSVMClassifier.epo'

classifier.Save, URI=classifierURI

; Evaluate the result

confusionMatrix = ENVIEvaluateClassifier(splitExamples[1], classifier)

; Print the confusion matrix

Print, confusionMatrix.Confusion_Matrix

; Print the column totals

columnTotals = confusionMatrix.ColumnTotals()

FOR i=0, (outExamples.NCLASSES)-1 DO $

  Print, 'Ground truth total for ', $

  outExamples.CLASS_NAMES[i],': ', $

  columnTotals[i]

; Print the row totals

rowTotals = confusionMatrix.RowTotals()

FOR i=0, (outExamples.NCLASSES)-1 DO $

  Print, 'Predicted total for ', $

  outExamples.CLASS_NAMES[i],': ', $

  rowTotals[i]

; Print the accuracy metrics

accuracy = confusionMatrix.Accuracy()

Print, 'Overall accuracy: ', accuracy

kappa = confusionMatrix.KappaCoefficient()

Print, 'Kappa coefficient: ', kappa

commissionError = confusionMatrix.CommissionError()

Print, 'Error of commission: ', commissionError

omissionError = confusionMatrix.OmissionError()

Print, 'Error of omission: ', omissionError

F1 = confusionMatrix.F1()

Print, 'F1 value: ', F1

precision = confusionMatrix.Precision()

Print, 'Precision: ', precision

producerAccuracy = confusionMatrix.ProducerAccuracy()

Print, 'Producer accuracy: ', producerAccuracy

recall = confusionMatrix.Recall()

Print, 'Recall: ', recall

userAccuracy = confusionMatrix.UserAccuracy()

Print, 'User accuracy: ', userAccuracy

; Normalize the data

normalizedRaster = ENVIGainOffsetRaster(raster, gain, offset)

; Classify the normalized attribute image

classRaster = ENVIClassifyRaster(normalizedRaster, classifier)

; Display the result

view = e.GetView()

layer = view.CreateLayer(raster)

layer2 = view.CreateLayer(classRaster)

view.Zoom, /FULL_EXTENT

END

This code saves the trained classifier to a binary file named TrainedSVMClassifier.epo. You can apply this classifier to other similar datasets that have the same attributes and data representation. The following example shows how to do this.

1. Get the sample data files (AttributeImage2.dat and TrainingDataROIs.xml) from our website or from the ENVI Resource DVD in the classification directory. Copy this file to your local drive.
2. Copy and paste the code below into the IDL Editor.
3. Update the following lines to include the full path where you saved the sample data:
   • file = 'AttributeImage2.dat'
   • ROIfile = 'TrainingDataROIs.xml'
4. Update the following line to include the full path where you saved the trained classifier:
   • trainedClassifierURI = 'TrainedSVMClassifier.epo'
5. Save the file as ApplySVMTrainedClassifier.pro.
6. Compile and run the code.

Tip: The classification image in this example overlaps with the image in the first example. As long as the ENVI application is still open from the first example, you can compare the results where they overlap. The classification in the overlapping region will not be exactly the same because the attribute values are different in each image; however, the results should be similar.

PRO ApplySVMTrainedClassifier

COMPILE_OPT IDL2

; Start the application

e = ENVI()

; Open an attribute raster to classify

file = 'AttributeImage2.dat'

raster = e.OpenRaster(file)

; Normalize the attribute image using the

; previously recorded gains and offsets

gain = [0.68837909, 0.039032014, 1.3670539, $

  1.2792631, 1.2774655, 1.0320982]

offset = [-0.14015123, -0.13057724, -0.12848703, $

  -0.16770373, -0.18784684, -0.28932331]

normalizedRaster = ENVIGainOffsetRaster(raster, gain, offset)

; Get the trained classifier. This is a persistable object

; in the form of an IDL save file that needs to be restored

; from disk

trainedClassifierURI = 'TrainedSVMClassifier.epo'

trainedClassifier = ENVIRestoreObject(trainedClassifierURI)

; Classify a different attribute image

classRaster = ENVIClassifyRaster(normalizedRaster, trainedClassifier)

; Get the classification raster metadata

numClasses = classRaster.Metadata['Classes']

classNames = classRaster.Metadata['Class Names']

; Display the result

view = e.GetView()

layer = view.CreateLayer(raster)

layer2 = view.CreateLayer(classRaster)

view.Zoom, /FULL_EXTENT

END

You can optionally use the ENVICalculateConfusionMatrixFromRaster function to evaluate the classifier. This function takes in a classification raster and training ROIs, and it calculates a confusion matrix and accuracy metrics. Add the following lines of code to the previous code example, after classifying the attribute image and before displaying the results. Update the ROIfile variable to include the full path where you saved the sample data:

; Open an ROI file that will be used to 

; evaluate the classifier

ROIfile = 'TrainingDataROIs.xml'

rois = e.OpenROI(ROIfile)

; Evaluate the classifier

confusionMatrix = ENVICalculateConfusionMatrixFromRaster( $

  classRaster, rois)

; Print the confusion matrix

Print, confusionMatrix.Confusion_Matrix

; Print the column totals

columnTotals = confusionMatrix.ColumnTotals()

FOR i=0, (numClasses)-1 DO $

  Print, 'Ground truth total for ', $

  classNames[i],': ', $

  columnTotals[i]

; Print the row totals

rowTotals = confusionMatrix.RowTotals()

FOR i=0, (numClasses)-1 DO $

  Print, 'Predicted total for ', $

  classNames[i],': ', $

  rowTotals[i]

; Print the accuracy metrics

accuracy = confusionMatrix.Accuracy()

Print, 'Overall accuracy: ', accuracy

kappa = confusionMatrix.KappaCoefficient()

Print, 'Kappa coefficient: ', kappa

commissionError = confusionMatrix.CommissionError()

Print, 'Error of commission: ', commissionError

omissionError = confusionMatrix.OmissionError()

Print, 'Error of omission: ', omissionError

F1 = confusionMatrix.F1()

Print, 'F1 value: ', F1

precision = confusionMatrix.Precision()

Print, 'Precision: ', precision

producerAccuracy = confusionMatrix.ProducerAccuracy()

Print, 'Producer accuracy: ', producerAccuracy

recall = confusionMatrix.Recall()

Print, 'Recall: ', recall

userAccuracy = confusionMatrix.UserAccuracy()

Print, 'User accuracy: ', userAccuracy