Dataset used in this work is from USGS Landsat-8 satellite. See the source here https://earthexplorer.usgs.gov/

Change-Detection

This code calculates difference from NDTI images. It can be used for change detection on the images with different dates. It can be easily modified for other indices.

Libraries that are used in jupyter notebook on python 3.8.5

Using a conda enviroment with jupyternotebook for this code could be helpful.

• rasterio 1.1.8
• numpy 1.19.4
• matplotlib 3.3.3
• gdal 3.1.4
• scipy 1.5.3

Input Bands for landsat8

• Band6 (SWIR1)
• Band7 (SWIR2)

Output Images

• NDTI Images
• NDTI difference
• NDTI threshold Image
• NDTI threshold removed noise Image

2019 RGB Image Kocaeli	2014 RGB Image Kocaeli

2019 NDTI Image Kocaeli	2014 NDTI Image Kocaeli

NDTI difference Threshold square image	After 3x3 Mode Filter

References

• Separating Built-Up Areas article (Osgouei et al. 2018)
• Change Detection between NDVI Images by Thales Sehn Körting
• NDVI Index calculation by Hatari Labs