Efficient Methods of Inactive Regions Padding for Segmented Sphere Projection (SSP) of 360 Video

Yong-Uk YOON  Yong-Jo AHN  Donggyu SIM  Jae-Gon KIM  

IEICE TRANSACTIONS on Information and Systems   Vol.E101-D   No.11   pp.2836-2839
Publication Date: 2018/11/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2018EDL8069
Type of Manuscript: LETTER
Category: Image Processing and Video Processing
JVET,  360 videos,  projection,  visual artifacts,  padding,  

Full Text: PDF(756.7KB)
>>Buy this Article

In this letter, methods of inactive regions padding for Segmented Sphere Projection (SSP) of 360 videos are proposed. A 360 video is projected onto a 2D plane to be coded with diverse projection formats. Some projection formats have inactive regions in the converted 2D plane such as SSP. The inactive regions may cause visual artifacts as well as coding efficiency decrease due to discontinuous boundaries between active and inactive regions. In this letter, to improve coding efficiency and reduce visual artifacts, the inactive regions are padded by using two types of adjacent pixels in either rectangular-face or circle-face boundaries. By padding the inactive regions with the highly correlated adjacent pixels, the discontinuities between active and inactive regions are reduced. The experimental results show that, in terms of end-to-end Weighted to Spherically uniform PSNR (WS-PSNR), the proposed methods achieve 0.3% BD-rate reduction over the existing padding method for SSP. In addition, the visual artifacts along the borders between discontinuous faces are noticeably reduced.