International Organization for Standardization Organisation Internationale De Normalisation Iso/iec Jtc1/sc29/wg11 Coding of Moving Pictures and Associated Audio Information

An object–based layered codec is submitted for formal testing of object scalability and improved coding efficiency at very low bit rates using class–A sequences. The codec consists of 3 layers. In layer 1, only colour parameters are transmitted. In layer 2, additionally motion parameters are transmitted. In layer 3, additionally to colour and motion parameters shape parameters are transmitted. In order to achieve the highest possible coding efficiency in compression applications, always the most efficient layer is selected according to appropriate switching criteria. In scalability applications, only layer 3 is used. As layers 1 and 2, TMN 5 has been incorporated with its intra– and interframe modes. In layer 3, object–based analysis– synthesis coding as derived from [7] is performed. It consists of 5 main components: image analysis, parameter coding, parameter decoding, image synthesis, and a memory for the object parameters. By the image analysis, the images of a sequence are decomposed into arbitrarily shaped moving objects each described by three sets of parameters defining the shape, the motion, and the colour of the object. The colour parameters consist of the luminance and chrominance values of the object surface. Goal of the approach implemented as layer 3 is to synthesize an image area as large as possible from colour parameters transmitted with previous images using only shape and motion parameters of the current image. Coding of colour parameters is concentrated on a small area, thus the available number of bits per coded picture element is relatively high guaranteeing a good quality in these areas. By the incorporation of TMN 5 as layers 1 and 2, it is guaranteed that the complete coder can not perform less efficiently than TMN 5. Part of the work was performed within the scope of the European COST 211ter project.