Stabilizing 360° Videos

Abstract

In this paper, we concentrate on stabilizing the 360° videos which when captured instantly causes unnecessary shakiness result in cyber sickness to the spectator. 360° videos are video recordings where we get the complete view around us at the same time. We implemented our algorithm in Matlab version R2016b. In our algorithm, we first convert the unstabilized 360° video which is usually represented in equirectangular to cubemap. Then we extract Speeded Up Robust Features (SURF) features from each of the faces and estimate a set of outliers and inliers using RANSAC algorithm. When we apply the warping technique in each frame, we can stabilize the 360° video.

Keywords: Cubemap, feature extraction, SURF features, RANSAC, warping

Cite this Article: Farzana Nazim, Krishna P, Reshma Chandran C R, Afsal S. Stabilizing 360° Videos. International Journal of Algorithms Design and Analysis. 2019; 5(2); 52–58p.