A semantic concept for the mapping of low-level analysis data to high-level scene descriptions

Along with the growing need for security, an increasing amount of surveillance content is created. To enable quick and reliable searches on the output of hundreds or thousands of surveillance sensors installed at a single facility, it is indispensable to index the content in advance. For this purpose, the concept of Smart Indexing and Retrieval (SIR) enables cost efficient searches by generating high-level meta data. As creating this meta data manually becomes more and more difficult to handle in reasonable time at reasonable costs, this has to be done automatically from the low-level features extracted by content analysis algorithms. Whereas formerly proposed concepts have been strongly application dependent in this paper a generic concept for mapping the results of the low-level content analysis data to semantic event descriptions is introduced. The constituting elements of this approach and their underlying concepts as well as an introduction to their application are presented. Finally, an example description of a scene is given and the applicability of the approach to great variety of application scenarios is demonstrated.