Unsupervised Classification for Illegal Building Monitoring

Abstract

In 2013 the Ministry of Construction and Physical Planning has brought an act by which all illegally built objects must be legalized. To this date almost 75% legalization request has been solved. It is expected that by the end of 2019 all of the illegally built objects will be legalized. In order to prevent further construction of illegal objects the Ministry of Construction and Physical Planning is seeking a way to easily detect start of illegal construction. Since the Copernicus satellite images are available free of charge and with resolution of 10m it should be possible to detect mentioned objects. This paper will provide analysis of Copernicus Sentinel 2A imagery for such use based on unsupervised classification using machine learning. If such procedure results in satisfying accuracy it will be proposed model for automation of the process for monitoring the illegal building construction based on Sentinel 2A imagery.