Metode și algoritmi de extracție și căutare de informații în serii temporale de imagini satelitare / Methods and algorithms for information retrieval and search in satellite image time series

The multitude of Earth observation missions generates data that can be used in applications that monitor natural resources or various physical phenomena that can affect human lives. Time series of satellite images contain a very large volume of useful information for the variety of possible applications, but for their processing, dedicated methods and software systems capable of working with very large volumes of remote sensing data are needed. The thesis aimed at developing software architectures for systems capable of extracting information from very large volumes of remote sensing data, identifying and integrating learning algorithms suitable for scenarios where there are few training examples, and creating software solutions capable of increasing both the level of use of remote sensing data by the scientific community, as well as the quality of the resulting information products. Two directions of time series analysis were approached, necessary to obtain the two categories of information mentioned above. Within the first direction, software tools and systems have been developed that implement computational optimizations and algorithms for the independent processing of remote sensing acquisitions. The second direction focused on the development of systems and software methods for the analysis of time series and the recognition of semantic evolution classes contained in remote sensing data. In order to facilitate the manipulation of very large volumes of data in order to extract the information contained, efficient software tools have been developed, dedicated to the generation of time series, indexing, searching, and interpretation of information extracted from time series of satellite images.