E1 216 : Computer Vision
2024
LECTURE 04: APPLICATIONS OF RADIOMETRY
In the last lecture we looked at the radiometry of image formation.
In this lecture we will look at the use of radiometry for reconstructing the shape of a surface.
Slides for
Lecture 04
Note that these slides also contain a bit of elementary linear least squares.
There are also a few slides of material on integrating normals to recover 3D surfaces which we will not discuss in class (not part of your syllabus for homeworks or exams).
Notes:
- Read Sections 2.1 and 2.2 of Practical Least-Squares for Computer Graphics. A number of books on numerical computation cover various methods for linear least squares estimate.
- The singular value decomposition (SVD) is a very important concept in linear algebra and serves as a very useful tool for solving many linear formulations of vision problems. A useful introduction to the singular value decomposition is Chapter 3 of Mathematical Methods for Robotics and Vision. For exams and assignments I shall assume that you understand the contents of this chapter.
- A very useful resource is the book Numerical Algorithms listed on this page.
- Suggested Reading: Why Gaussianity? deals with the ubiquity of the Gaussian distribution.