![Satellite_Image_Analysis[1]](https://image.slidesharecdn.com/c9c0f674-0e71-4359-8363-b1fedfeafaa0-161221063916/85/Satellite_Image_Analysis-1-2-320.jpg)
![Satellite_Image_Analysis[1]](https://image.slidesharecdn.com/c9c0f674-0e71-4359-8363-b1fedfeafaa0-161221063916/85/Satellite_Image_Analysis-1-20-320.jpg)
Satellite_Image_Analysis[1]
- 1. UNIVERSITY OF BAGAMOYO COLLEGE OF SCIENCE, INFORMATICS AND BUILT ENVIONMENT Bachelor of Science in Geoinformatics ADVANCED GIS SGI 2222 Analysis of Satellite Image of Mbulu BY USING SUPERVISED CLASSIFICATION Author: Joachim Nkende Yohana Registration No: UB010/0053/14 Supervisor: Mtalo E. G. Submission date: 17/06/2016
- 3. 1 STEP 1 SUPERVISED CLASSIFICATION IN GIS Supervised classification is the process of using training data to assign objects of unknown identity to one or more known features. The more features and training samples you select, the better the results from supervised classification. Training data consist of objects that you select as representative samples of known features. Supervised classification can be used to cluster pixels in a dataset into classes correspond- ing to user-defined training classes. This classification type requires that you select training areas for use as the basis for classification. Various comparison methods are then used to determine if a specific pixel qualifies as a class member. According to mbulu satellite images, the composition of three satellite images was done on R so as to allow easy image projections and combining them using stack and writeRaster functions. Semi Automatic- Classification plugin was downloaded , installed then loaded to Q GIS The R output which is the mbulu.tiff file was then loaded to Q GIS, followed by selecting the false color composite as RGB 3, 2, 1
- 4. 2 The Q GIS window could be now displayed as shown below; The first step in SCP was ROI creation, where by it was required to begin with creation of shapefiles
- 5. 3 Mbulu-Shapefile was created containing the following polygons: 1. Water-bodies 2. Forests 3. Shrubs 4. Grassland and 5. Bare-Soil SCP Classification could then start by highlighting and saving the ROI Signature list as and then click Export .With this project exported.xml was saved to working directory. The next step is to edit the reflectance in attribute tables in band1, band2 and band3 as 485, 560 and 660 respectively(Water-bodies1-lake1.csv, Forests2-Natural2.csv, Shrubs3- Natural3.csv, Grassland4-Natural4.csv and Bare-Soil5-Silt5.csv) followed by adding the
- 6. 4 highlighted signatures to spectral signature plot. There after it was required to select classifi- cation algorithm to maximum likelihood and adjust threshold to 100 then save Maximum likelihood classification assumes that the statistics for each class in each band are normally distributed and calculates the probability that a given pixel belongs to a specific class. Unless a probability threshold is selected, all pixels are classified. Each pixel is assigned to the class that has the highest probability (i.e., the maximum likelihood). Then highlighted signature list was added to spectral signature plot to view the reflectance
- 7. 5 The reflectance’s detail was observed also in signature details tool bar
- 8. 6 Classification preview scale was adjusted to 1000 to cover total area, then classification report was marked before clicking Perform Classification bar Output was saved as MbuluClassified.tif as shown below
- 9. 7 STEP 2 CREATING A MODEL IN IN QGIS The process below shows the basic steps which were used in creating a model; To begin , launch Q GIS software and open processing tool bar. Click on the Graphical modeler to continue.
- 10. 8 Since the Processing modeler dialog contains a left-hand panel and a main canvas. Selection of the Inputs tab in the left-hand panel was done ,following the dragging of + Raster layer to the canvas. Then after, selection of Algorithms tool bar was made to allow input of majority filters because the Majority Filter algorithm uses the Input raster as its input.
- 11. 9 The next step was to convert the output of majority filter to vector through the Polygonize (raster to vector) algorithm. After dragging the algorithm ‘Majority Filter’ then was used as the set function of converting raster to vector.
- 12. 10 The next step in the work flow was to query for a class value and create a new layer from the matching features by using the Extract by attribute algorithm by drag it the canvas. Selection of ‘Vectorized’ from algorithm ‘Polygonize (raster to vector) as the Input Layer was made followed by entering DN as the Selection attribute and 12 as the value. Then the output of this operation as it is the final result or output was named as vectorized class
- 13. 11 The Model name was named as vectorize and Group name as raster when saving the model Since the model was saved successfully, it was possible to test the model now. It was required to close the modeler and switch to the main QGIS window. The Layer button was used to Add Layer by attribute Add Raster Layer MbuluClassfied.tif raster layer was added
- 14. 12 A new created model in processing tool box named vectorize was used to run Mbulu- Classfied.tif raster layer as input Successively the raster was changed to vector, however further editing in the model could be made to specify input parameter which the user can change instead of using 12 as a default value.
- 15. 13 To add this, it was required to switch the Inputs tab and drag the + String to the model.Then enter the Parameter Name as Class. Enter 12 as the Default value. Then changing the Extract by attribute algorithm to use this input instead of the hard- coded value by using the Edit button next to the Extract by attribute box. Then the DN value was changed to class instead of 12
- 16. 14 As shown below Extract by attribute algorithm now uses 2 inputs, to view the output now we need to run the model STEP 3 GIS POLYGON SELECTION (OUTPUT) To display an image , it needs to edit the input value such as how it is shown below; First you need to select the classified image eg. MbuluClassified.tif
- 17. 15 Then edit the majority filter search mode to Circle and adjust Radius at least to 10, also it requires to enter output raster name. It also required to set Output-vector name
- 18. 16 Next step was to save the model and run. In order to overcome errors, first run the selected outputs in temporary file.
- 19. 17 Outputs will be many depending on the selected number of outputs, and will be displayed as follows;