A Kalman-filtering method for 3D camera motion estimation from image sequences

Abstract

In this paper, we describe a method for estimating and compensating 3D camera motion in image sequences for applications to the video coding system. To effectively estimate camera motion parameters (zoom, pan, tilt, swing, and focal length) from image sequences, we propose a new linear motion parameter equation and use the Kalman filtering method to solve it. Unlike the existing linear techniques, the proposed linear method accurately estimates the large rotation angles and the focal length. Experimental results show that the proposed method outperforms the conventional linear methods, especially for a large rotation angle.