Skip to content

Pixel based classification of satellite imagery - feature generation using Orfeo Toolbox, feature selection using Learning Vector Quantization, CLassification using Decision Tree, Neural Networks, Random Forests, KNN and Naive Bayes Classifier

License

Notifications You must be signed in to change notification settings

kkgadiraju/SatelliteImageClassification

Repository files navigation

SatelliteImageClassification

Pixel based classification of satellite imagery

  • sample training and testing points generated using Point Sampling plugin in QGIS
  • feature generation using Orfeo Toolbox
  • feature selection using Learning Vector Quantization
  • CLassification using Decision Tree, Neural Networks, Random Forests, KNN and Naive Bayes Classifier
  • Ensemble classifier for Flood Inundation Mapping - classifies a pixel as water if 2 or more than 2 of the above classifiers classify a pixel as water
  • Mode filter used to remove individually wrongly classified pixels
  • Classification accuracy to measure goodness of each model

Outcomes of the best classifier (Multi Layer Perceptron) are as shown below:

To compile and run SatelliteClassification.java, you need weka.jar that you can download from the Weka website.

Compile code:  javac -cp weka.jar SatelliteClassification.java  
Run code: java -cp weka.jar:. SatelliteClassification  "trainingFile" "testingFile" "classifiername"
  • order:

|clouds: white | |roads: yellow | |shadow: black | |urban: pink | |vegetation: green | |water: blue |

Original LANDSAT 8 Image during Flooding Multi Layer Perceptron Classification
Original LANDSAT 8 Image during Flooding Multi Layer Perceptron Classification
Ensemble Classifier: Water vs Everything, without filtering Ensemble Classifier: Water vs Everything, after mode filtering
Ensemble Classifier Ensemble Classifier - After Mode Filter

References:

About

Pixel based classification of satellite imagery - feature generation using Orfeo Toolbox, feature selection using Learning Vector Quantization, CLassification using Decision Tree, Neural Networks, Random Forests, KNN and Naive Bayes Classifier

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published