Computing Egomotion and Detecting Independent Motion from Image Motion Using Collinear Points

摘要

This paper presents and analyzes a new constraint (the Collinear Point Constraint) for use in computing egomotion and detecting independently moving objects from image motion. The constraint provides an elegant means of cancelling the rotation of the observer while simultaneously providing evidence for the location of the Translational Focus of Expansion. It also provides the minimal computation necessary for detecting independent motion.