Robust global motion estimation for video stabilization

Abstract

Global motion estimation is an important task for video stabilization. Since the global motion is heavily related to the motion of the background (BG), the moving foreground (FG) objects can cause inaccurate global motion estimation. In this paper, we propose a robust global motion estimation method for video stabilization. The proposed method iteratively finds the feature points (FPs) in the BG and estimates the global motion using the BG FPs. The experimental results show that the proposed method can estimate the global motion even when the moving FG objects occupy a large part of image.

Robust camera motion estimation in presence of large moving objects

“…Camera motion estimation currently supports a wide range of applications as video coding [1] [2], camera stabilization [3] [4][5] [6] or object segmentation [7][8] [9]. In most cases, the terms camera and dominant/global motion are used interchangeably and GME algorithms are assumed to extract the camera/background motion, even if this motion is not dominant.…”

“…In this paper we also follow this convention. GME algorithms can derive the camera motion either from the image intensity values (as phase correlation [6] [10] and differential based techniques [5] [8]) or from some previous estimation of local motion (as feature based [3] [9] and optical flow based techniques [7] [11]). But their suitability for a particular application mostly depends on their behavior with data contaminated by pseudo-outliers (data from alternative structures, i.e.…”