Monocular Vision — Segmentation in Additive Images
This chapter begins the presentation of the core material in the book. Here, we develop and apply a technique to separate additive images (i.e. images formed by the addition of a number of component images) into their additive components. We will assume that each component image is translating with some unknown and possibly zero velocity. Specifically, given a time sequence of a number of images, each formed by the addition of two distinct but perhaps identical images — distinct in the sense that the component images are moving with a different velocity — we will show how one can recover the two component images. In the next chapter, we will extend these results to the case where the images are not formed by additive superposition but by occluding superposition, i.e. the images comprise two objects or scenes, one occluding the other.
Unable to display preview. Download preview PDF.