Image Motion Recovery Using the Method of Total Least Squares

Abstract

Recovering image motion in the form of a velocity field is particularly useful in image analysis when the signal-to-noise ratio is low in individual frames of an image sequence. The velocity vector of a point on an image plane is the position change of the projection of an object point due to a change of viewing angle between the sensor and the object point. A change in the viewing angle can be caused by the velocity of the sensor, or the movement of the object point in the scene, or both. The velocity vector is thus an approximation to the velocity of the image point.