โ GIS & Remote Sensing learning pathGIS & Remote Sensing ยท Module 9 of 10
Image processing and classification
Turning images into information.
๐ Key ideas
- Pre-processing: radiometric and atmospheric correction, georeferencing and cloud masking.
- Band combinations (false colour composites) highlight vegetation, water or urban areas.
- NDVI = (NIR - Red) / (NIR + Red); values near +1 indicate dense healthy vegetation.
- Classification groups pixels into land cover classes: supervised (with training samples) or unsupervised (clustering).
- Accuracy assessment compares classified maps with ground truth using a confusion matrix.
๐ก Real-life example: Like sorting photos into albums: training the computer with examples of water, forest and city.
NDVINormalised Difference Vegetation Index
Supervised classificationClassification using training samples
GeoreferencingAligning an image to real-world coordinates
Confusion matrixTable comparing classified and true classes