Computing Intrinsic Images.

Low level modern computer vision is not domain dependent, but concentrates on problems that correspond to identifiable modules in the human visual system. Several theories have been proposed in the literature for the computation of shape from shading, shape from texture, retinal motion from spatiotemporal derivatives of the image intensity function and the like. The problems with the existing approach are basically the following: (1) The employed assumptions are very strong and so most of the algorithms fail when applied to real images. (2) Usually the constraints from the geometry and the physics of the problem are not enough to guarantee uniqueness of the computed parameters. (3) In most cases the resulting algorithms are not robust, in the sense that if there is a slight error in the input this results in a catastrophic error in the output. In this thesis the problem of machine vision is explored from its basics. A low level mathematical theory is presented for the unique robust computation of intrinsic parameters. The computational aspect of the theory envisages a cooperative highly parallel implementation, bringing in information from five different sources (shading, texture, motion, contour and stereo), to resolve ambiguities and ensure uniqueness and stability of the intrinsic parameters. The problems of shape from texture, shape from shading and motion, visual motion analysis and shape and motion from contour are analyzed in detail.