SAR sensors measurements for environmental classification: Machine learning-based performances

Artificial intelligence, in particular a supervised and unsupervised machine learning approach, has been becoming an interest in the field of measurement and instrumentation. Many problems of classification can be faced by a machine learning approach. We know machine learning is a broad area of artificial intelligence that comprises some other lines of research and activities such as deep learning. Synthetic aperture radar (SAR) measurements by means of its sensors are of great interest in environmental monitoring, in particular in land classification. This paper presents findings related to measurements and characterization through land classification of an environmentally sensitive area in Italy over two different time periods in order to assess changing parameters. A deep learning algorithm has been designed and implemented, and a comparison has been established with a spectral density approach.