Gyroscope Aided Video Stabilization Using Nonlinear Regression on Special Orthogonal Group

Abstract

This paper presents a novel approach for gyroscope aided video stabilization. With the raw 3D rotational motion captured by a gyroscope, it is then smoothed through nonlinear regression directly on the Special Orthogonal Group. Instead of solving a variational problem, the regression problem is discretized with finite forward difference, which makes it an optimization problem on manifold. We derive a quadratic approximation of the objective function using Lie algebra. To address the black border problem caused by smoothing, we model it as linear inequality constraints. The resulting quadratic programming problem can be efficiently solved. Experiments on synthetic data and real video sequences demonstrate that our method performs better than the compared method in terms of motion smoothing and video stabilization.