Open Collections

Slant from texture : computational methods for recovering surface slant from images of textured scenes

University of British Columbia

This thesis examines the problem of computationally recovering or determining the slant of a surface from an image of that surface. Images are restricted to those of planar surfaces produced by orthographic projection. This thesis is concerned only with those cues obtainable from the image texture. These cues arise primarily due to the foreshortening property of orthographic projection. Texture measures have typically been partitioned into three classes: statistical approaches, micro-structural approaches, and macro-structural approaches. In this thesis, measures from each of these classes are used to develop algorithms capable of detecting surface orientation. It is concluded that these three classes are not distinct and, indeed, are artificially rendered by the prevailing definition of texture. A new definition involving nested structures is suggested.

Text

2010-03-24

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/22461

Computer Science

University of British Columbia

Graduate