A universal k-tree model for content-based multimedia retrieval

In this paper, we propose a unified model for indexing and retrieving multimedia data using characteristic features and multiresolution processing. Each multimedia datatype, such as audio and video, is represented as a k-dimensional signal in the spatio-temporal domain. A k-dimensional signal is transformed into characteristic features (contents) and these features are stored in a hierarchical multidimensional structure, called the k-tree. With this approach, both the retrieval accuracy and the response time can be dynamically adapted. We investigate the multimedia data representation and classify types of data retrieval based on such a universal ktree representation of the characteristic features. We propose multiresolution approaches to content-based multimedia search and retrieval.