What is it about?
Next frame prediction is a popular research direction, but existing works about frame prediction are all based on 2D plane video which can not predict frames in high resolution like 360-degree video, Furthermore, they all use complicated networks that costs a lot of time designing and training. However, we predict a future frame of 360-video only by rotating the sphere. This simple method can predict shape deformation accurately, compared with training complicated networks. By inserting the predicted frame in the reference list, we can improve the 360 video compression efficiency.
Featured Image
Photo by Bernard Hermant on Unsplash
Why is it important?
Shape deformation is the most important characteristic of 360-degree video. It also brings many difficulties in coding because it is hard to search for a similar block during motion estimation. We come up with this idea that predicts the shape deformation of the current encoding frame by a convenient but efficient method. In this way, our scheme can improve both BD-gain and accuracy of motion and shape deformation towards 360 standards videos.
Perspectives
4Writing this article was a great pleasure as it gives me a lot of motivation to read many excellent articles about 360 video coding which inspires me a lot. I come up with this idea by considering the generation of shape deformation that it happens during objects move to the different latitude of the sphere. I hope this “rotating the sphere” method can be useful in applications of 360-degree video compression more or less.
chunyu lin
Beijing Jiaotong University
Read the Original
This page is a summary of: 360 video compression based on sphere-rotated frame prediction, IET Image Processing, June 2020, the Institution of Engineering and Technology (the IET),
DOI: 10.1049/iet-ipr.2019.1663.
You can read the full text:
Contributors
The following have contributed to this page
