Towards a modeling of video grammar based on a priori knowledge for the optimization of the audiovisual documents structuring

The emergence of the TV channels number and the appearance of the Internet led to an exponential growth of the audiovisual documents. Several works were proposed in the literature for automatic video segmentation. However, the effectiveness of these works depends on video type. So, for a better quality of the segmentation, it is necessary to consider a priori knowledge concerning video types. Besides, TV channels use a set of rules during the production and the streaming of their programs such as the respect of the graphic charter, the recurring of studios... The exploitation of these rules, as a priori knowledge for the temporal and/or semantic structuring of the video documents, would be then advantageous for the quality of the indexing and the performance of the multimedia systems. In this context, we propose a method for the extraction and the formalization of descriptors based on a priori knowledge in order to conceive a video grammar model. This model will be used thereafter for a multitude of interesting applications like automatic structuring of TV streams or TV news segmentation. Video grammar; video grammar modeling; multimedia descriptors; video indexing.