Efficient Video Coding Considering a Video as a 3D Data Cube

All existing video coding standards consider a video as a temporal (along T-axis) collection of two dimensional pictures (formed by XY axes) and compress them by exploiting spatial and temporal redundancy in the pictures. A recent optimal compression plane (OCP) determination technique shows that better compression can be achieved by relaxing the physical meaning of axes by exploring information redundancy in a fuller extent where a video is considered as a 3D data cube. Spatial and temporal dimensions are determined based on the statistical redundancy along each axis. Treating a video as a 3D data cube revolutionizes the traditional video features such as background, motion, object, zooming, panning, etc. In this paper we apply dynamic background modeling to the OCP plane to exploit the newly generated background in the video for further improving the coding performance. The experimental results reveal that the proposed approach outperforms the existing state-of-the-art OCP technique as well as the H.264 video coding standard.