Контекстное моделирование объектов земной поверхности с целью получения метаданных о содержании изображений

Development of technologies for remote monitoring of Earth surface prompts to look at new approaches for creating of metadata for digital image catalogues and archives. The size of total digital data continues to grow at tremendous pace, mainly due to the start of new generation of satellites (WorldView-1, TerraSAR-X, WorldView-2), which bring terabytes of information in every new image. Digital catalogs and archives of spatial information require entirely new approaches to creating metadata in connection with the search capabilities on the contents of visual objects and the need to establish the level of spatial detail for the stored information. We propose a model within a new conceptual framework, which overcomes the differences between the classes of natural objects of the Earth's surface in terms of their spatial and spectral representation. The model provides new opportunities in highlighting the meaningful visual structure in a uniform way, formalization for assessments of environmental specialists, as well as automatic annotations for satellites' images.