Towards a new approach for video database semantic annotation

Video database annotation has become the area field of interest recently, where several methods and techniques emerge. Regarding manual annotation as seriously time-consuming and automatic ones as large-scale of video unsatisfied, our topic of interest belongs to the semi-automatic annotation, where a little set of video is annotated manually and this will be propagated to the entire databases. We try to consider annotation able to exploit unconsidered concepts to help for improving annotation and XML performance. Our approach takes the low level descriptors´ MPEG7 file added to the ontology´s lexicon as input, and defines concepts of high level as output. Hence, an annotated video database is created.