HomeProjectsInformation and Communication TechnologiesEfficient Mapping Algorithms For 360-Degree Videos

Efficient Mapping Algorithms For 360-Degree Videos

Project Quick Facts

Principal Investigator

  • Prof. WONG Tien-Tsin

    Department of Computer Science and Engineering

  • Funding

    Research Grants Council of Hong Kong

Mapping solutions for enhancing the quality and the efficiency of 360-degree video.

Recently, Virtual Reality (VR) has become the basis of the future I.T. industry. Different from flat displays that display flat videos, VR devices are able to display 360-degree scenes that are more close to human vision. To display VR videos (or 360-degree videos), spherical mapping is usually used to display the stored

data in a 360-degree view. However, classic spherical mapping is inefficient in generating high-quality (e.g. 2k/4k image resolution) 360-degree images/videos. Blurry artifacts are quite common when the images/videos are played on high-resolution displays. To efficiently store and map 360-degree images/videos with high quality, we propose several novel mapping schemes to enhance the quality and the efficiency of 360-degree video mapping, including fast isocube (novel six-face) spherical mapping, rhombic dodecahedron mapping, unicube spherical mapping for dynamic environment mapping, and HEALPix spherical mapping.

Comparison between existing cubemap and our proposed isocube
Rhombic dodecahedron mapping
360-degree video frame
360-degree cameras

Uniqueness and Competitive Advantages:

  • Our method can benefit anything that involves spherical mapping and playback, which is a key feature in 360-degree video streaming, VR networking and digital entertainment
  • Our mapping schemes can be applied directly to all existing VR videos/360-degree videos on existing hardware platforms without requiring any further effort to build new hardware platforms i.e. no extra cost
  • Our methods can greatly improve the visual quality of the mapped 360-degree images/videos while remaining efficient. Our method achieves the maximum efficiency with GPUs while maintaining high fidelity and dynamic effects

Do you like our project?