A probabilistic adaptive algorithm for constructing hierarchical meshes

Recent advancement in the field of video communication requires an efficient video compression technique which can transmit a good quality video with very low bandwidth. Mesh based video compression techniques are often used to achieve this goal. A mesh is formed on the boundary of an object inside a video frame and nodes of the mesh are transmitted to describe the motion of the object. Motion vectors at the nodes decide on the coarse-to-fine hierarchy of a mesh. Sometimes this technique fails to construct a proper hierarchical mesh for an object and cannot capture its motion properly. This paper introduces a new probabilistic approach to construct a hierarchical mesh which can describe the motion of an object with fare accuracy without knowing the proper shape of the object. This algorithm can also sense the amount of motion and select the reference frames accordingly in proper places so that minimum information is required to be transmitted without compromising on the quality of the video.