A segmentation-based motion-compensated scheme for low-rate video coding

A system for coding video sequences at a low rate is presented. The system is very effective in coding sequences where motion is restricted to a not large area of each frame. Nevertheless knowledge based techniques are not used for not reducing the generalization power of the coding method. Techniques based on segmentation algorithms are used to encode both the displacement vector field and the prediction error image. Such techniques permit one to effectively distribute coding information in those areas of the image where it is required. Great care is given to a good estimation of motion of the scene objects: this is to reduce the amount of information to be coded in the prediction error image