Adaptive hierarchical grid motion estimation algorithms for video coding application at low bit-rate communication

The problem of reducing the temporal redundancy between consecutive frames of a video sequence in the context of videoconferencing application at low bit-rate transmission is considered. To better model the different types of motion, the block matching algorithm implemented in most of the current video coders is replaced by 2D adaptive rectangular mesh algorithms. It is through the deformation mesh that motions of objects are described. To predict the mesh objects in accordance with the appropriate coding modes selected type P or B, two classes of motion estimation algorithms are developed. The relevant information to be efficiently encoded such as the positions of nodes, their connections and errors of motion compensation is studied. The implementation of our algorithms in a complete coding scheme offers good performance compared to H.264/AV C at low bit-rate transmission.