Metadata-Oriented Concept-Based Image Retrieval for Forest Fire Video Surveillance System

摘要

In this paper we analyse and discuss two different approaches for archival image metadata storage and retrieval in forest fire video surveillance system. Video surveillance is an effective approach for the protection of valuable assets, such as buildings, work facilities and open areas. Forest fire video surveillance systems are used for forest monitoring since they allow rapid reaction in the case of forest fires. Video surveillance systems provide their users with real time images, but there is also a demand for storage of those images for event reconstruction processes. Each image in an archive has metadata associated with it. In this paper we compare two standard approaches for metadata storage: encoding metadata in images, and storing metadata in a separate database. We discuss advantages and disadvantages of these approaches from the aspect of user experience, and in terms of ease of image retrieval, portability, and search speed.