Object-based supervised classification

From AWF-Wiki
Revision as of 21:46, 19 June 2021 by Hfuchs (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Contents

Object-based image analysis (OBIA) with QGIS and OTB processing plugin

Segmentation

  • In the search engine of Processing Toolbox, type segmentation and double click Segmentation.
  • Select the input image: a multispectral Sentinel-2 image (data type uint16).
  • Set Segmentation algorithm to meanshift
  • The Range radius value can be set to 600. The optimal value depends on datatype dynamic range of the input image and requires experimental trials for the specific classifcation objectives.
  • Set Minimum Region size (in pixels) to 16.
  • Processing mode Vector
  • Set the Mask image to blank (top of drop-down list).
  • Check 8-neighborhood connectivity on.
  • The Minimum object size (in pixels) can be set to 16 depending on minimum mapping size.
  • Name the Output vector file e.g. segments_meanshift.shp. Extension should be .shp in this module).
  • Click Run.

Qgis otb segmentation.png

  • Evaluate the segmentation results: Load the output vector file segments_meanshift.shp into QGIS on top of the image Subset_S2A_MSIL2A_20170619T_Mul.tif

Mark the vector layer in the Qgis Layers window. Right click Properties --> Symbology --> Simple Fill, Fill Style: No Brush and Stroke color:white.

Feature extraction

In the search engine of Processing Toolbox, type zonalstats and open ZonalStatistics under Image Manipulation of OTB.

  • Select the Input image: a multispectral Sentinel-2 image.
  • Background value to ignore: 65535
  • For the Input vector data select a vector file with segments (output from Segmentation)
  • File name for the output vector data: segments_stats.gpkg.
  • Click Run.

Otb zonalstats.png

Training phase

  • In the search engine of Processing Toolbox, type Train and double click TrainVectorClassifier.
  • In the Input Vector Data List do not select a file from the list which is already loaded in the QGIS Viewer. There is currently a bug in QGIS 3.16 which leads to failure during file import. Instead please select a vector file clicking Qgis add file.png and browse directly to the file containing training area polygons in format GPKG or SHP e.g. lucc_training_obia.gpkg.
  • Output model filename is svm_obia.model
  • In the field Field names for training features copy and paste
 "mean_2 stdev_0 mean_9 mean_7 mean_0" 
  • This is one of many variable sets as a result of a feature selection procedure.
  • The name of Field containing the class id for supervision" is C_ID.
  • Classifier to use for training: libsvm
  • SVM Kernel Type: linear
  • SVM Model Type: csvc
  • Click Parameters optimizationON.
  • Click Run.

Qgis otb trainvector.png

info.png Info
For more detailed information on the SVM algorithm visit the LibSVM website

Classification phase

  • In the search engine of Processing Toolbox, type VectorClass and double click VectorClassifier.
  • In the Input Vector Data do not select a file from the list which is already loaded in the QGIS Viewer. There is currently a bug in QGIS 3.16 which leads to failure during file import. Instead please select a vector file clicking Qgis add file.png and browse directly to the file containing segments and features for the whole image (result of Feature extaction)training area polygons in format GPKG segments_meanshift_zonal.gpkg.
  • Name of the input model file is svm_obia.model.
  • Output field containing the class is C_ID
  • Copy and paste into the field Field names to be calculated have to be the same features for prediction as were defined before in the TrainVectorClassifier module:
 "mean_2 stdev_0 mean_9 mean_7 mean_0" 
  • Output vector Data file is lucc_classified_obia.gpkg.

Run.

Qgis otb vectorclassifier.png

Load the output vector file manually into QGIS and apply the same QGIS style used for the training data. Layer --> Layer properties --> Symbology > Style --> Load style....

Personal tools
Namespaces

Variants
Actions
Navigation
Development
Toolbox
Print/export