Semantic Analysis of Satellite Image Time Series
Corneliu Octavian Dumitru, Mihai P. Datcu · 2021
Satellite image time series (SITS) are high-complexity multidimensional signals represented in space, time and spectral domain. This chapter explains the need for semantics in earth observation (EO) SITS, and describes the similarities metrics. The angular distance is one of the most robust similarity measures at risk of changing due to radiometric correction inaccuracy and atmospheric effects. The chapter presents typical feature methods used for the SITS, and contains the classification methods for SITS. It proposes a number of algorithms developed and tested during different projects for different SITS acquired by the multispectral and synthetic aperture radar sensors. SITS are an essential asset in multidisciplinary EO applications and valorization domains. SITS semantics are important in the communication of meaning from the observed Earth dynamic process to the applications and valorization services.