Method of detecting the relative depth between two objects in a scene from a pair of images taken at different views

摘要

The method includes the following stages: - create a spatial segmentation of one of the images, denote it as the 'reference image'; - provide a local estimate of the movement between the two images, the resultant vector field is called 'local movement' ; - determine an estimate of errors along the movement fronts; - analyse the location of the movement errors along the movement fronts so as to concluded that a region A is in front of a region B when the movement error estimation along the front is principally found localised from the side B of this front.