360 Video Compression with an Advanced Spherical Motion Model and Local Padding PROJECT TITLE : Advanced Spherical Motion Model and Local Padding for 360 Video Compression ABSTRACT: The geometry distortion and the face boundary discontinuity are two of the key issues in 360Á video compression because of projection distortions. Either equirectangular projection (ERP) or polyhedron projection has advantages and disadvantages. A polyhedron projection's face boundary discontinuity is more pronounced than ERP's geometry distortion. Motion correction and compression efficiency will suffer as a result of these two distortions. Efficiencies in coding can be improved by integrating these two challenges into a single framework, as this research shows. There are two major contributions to the framework proposed here. The first step is to create an unifying advanced spherical motion model to deal with the geometry distortion of diverse 360Á video projection formats. Each projection format can be given a unique solution by integrating the multiple projection formats and the sphere into the unified framework. To address the issue of face border discontinuity between adjacent faces in 360Á video, we offer a local 3D padding solution. It is possible to apply the local 3D padding method to a variety of projection formats by adjusting the angles between adjacent faces. It is possible to combine these two approaches for improved rate-distortion performance in a single framework. New video coding standard high-efficiency video coding can be easily integrated with the suggested framework Experiments have shown that implementing the recommended coding methods can save large amounts of bitrate above and beyond what is currently possible. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest Spatiotemporal VLAD with an Action-Stage Emphasis for Video Action Recognition Single Image Dehazing With Atmospheric Illumination Prior AIPNet Image-to-Image