Contextual Geospatial Picture Understanding, Management and Visualization

We describe SPIM (Spatial Image Management), a framework and a system that allow identification and visualization of the semantics of distant objects in nature, which exist in pictures taken by mobile devices. Semantic objects in nature are represented on 3D maps and tagged with semantic and spatial descriptors. The system functionality creates, on top of pictures, a layer that allows semantic user browsing in augmented geospatial images, semantic object identification in pictures, object interaction, contextual map visualization, and interlinking of information in digital libraries and the web. We present requirements of semantic geospatial descriptions, system functionality, matching algorithms and experimental results.