Edge Aware Commonality Modeling based Reference Frame for 360 Degree Video Coding

Video coding algorithms tries to model the significant commonality that exists within a video sequence. This role is even more critical in 360-degree video sequences given the associated enormous amount of data that need be stored and communicated. Moreover, in such sequences the captured video signal from omnidirectional cameras are projected onto a plane; hence they exhibit different characteristics compared to traditional video frames. Therefore the conventional block-based and translational motion modeling approach employed by modern video coding standards, such as HEVC, may not provide an efficient compression of 360-degree video data. The edge position difference (EPD) measure based motion modeling (EPD-MM) has shown good motion compensation capabilities for traditional video sequences. The EPD-MM technique is underpinned by the fact that from one frame to the next one, edges map to edges and such mapping can be captured by an appropriate motion model. Since the 360-degree frames contain significant edge information, in this paper, for motion compensation an edge aware commonality modeling technique is adopted. In particular, the EPD-MM based motion compensated prediction of the current 360-degree frame is generated using the already coded reference frame. Experimental results show that if this predicted frame is used as an additional reference frame, bit rate savings can be obtained over an anchor HEVC encoder.