Set of visual feature descriptors and their combination in a low-level description scheme

This paper reports about descriptors for basic visual information features, which have been developed in the context of the forthcoming MPEG-7 standard. The four basic features supported are color, texture, shape and motion. A search engine system has been developed which supports combinations of basic feature descriptors in a low-level description scheme for similarity-based retrieval of visual (image and video) data. All basic descriptors have been developed so that invariance against common transformations of visual material, e.g. filtering, contrast/color manipulation, resizing, etc., is achieved, and that they are fitted to human perception properties. Furthermore, descriptors have been designed to allow fast, coarse-to-fine search procedures. Elements described in this contribution have been proposed for the MPEG-7 standard. They are currently either included in the MPEG-7 Experimentation Model (XM) or investigated within core experiments, which are performed duri ng the standard's development.