Abstract

Multiple view geometry is increasingly being used in machine vision. In the first part of this two part paper, we present a mathematical background of the multiple view geometry that is normally used in X-ray computer vision. The article describes an explicit model which relates the three dimensional coordinates of an object to the two dimensional coordinates of the digital X-ray image pixel, the geometric and algebraic constraints between two, three and four X-ray images taken at different projections of the object and the problem of three dimensional reconstruction from n views. Part II, to appear in the December issue, will present the application of multiple view geometry to X-ray testing and describe applications which use the theory.