Photometric stereo is a computer vision method of analyzing and detailing the contour and reflectivity of a surface in 3D (three-dimensional) space. The method involves shining an external light source on that surface, moving the light and gathering multiple images based on the resulting illumination scenarios.
Photometric stereo has diverse applications, most notably facial recognition, industrial product quality assurance / quality control, and analyzing the surfaces of celestial objects, notably the moon. Unlike a 3D scan or hologram, a photometric stereo image allows the observer to remain fixed; only the location of the light source changes. Photometric stereo determines the orientations of vectors (rays in space) that run normal (perpendicular) to a surface at multiple points. As the number of points and normal vectors increases, the quality of the resulting stereo map improves.
The mathematics of photometric stereo include advanced calculus, vector analysis, and differential equations, but the basic concept is easy to imagine. If you photograph an intricate statue from a single vantage point at a certain time of day (say, noon) in direct sunlight, then anyone who looks at the photo will get a fair idea of the statue's contour. However, if you photograph that same statue from that same vantage point at hourly intervals throughout a long, sunny summer day, anyone who sees the collection of photos will get a better idea of the statue's contour because the sun will have illuminated it from several different angles. In photometric stereo, a specialized computer program would combine images resulting from hundreds, thousands or millions of light-source orientations, coming from all possible directions in 3D space, to produce a precision mathematical image of the statue's surface -- but only as seen from a single reference frame.
The concept of photometric stereo is not new; Robert J. Woodham of the University of British Columbia introduced the idea in 1980. Since that time, the advancement of computer technology has vastly improved the precision and expanded the applicability of photometric stereo processes.
Continue reading about photometric stereo:
Photometric Stereo: Introduction
A Photometric Stereo Approach to Face Recognition