The course will give a basic knowledge on 3D-Computer Vision methods for extracting information on depth, shape, and motion in a scene from one, two, or several digital images. By analyzing the image formation process will shall see which parameters that can be controlled to facilitate the extraction. If quantitative measures is to be extracted a calibration process is necessary. The course will cover the following issues:
A essential element in the course will be two mandatory exercises. These will involve experimental work at the DIKU image laboratory where a high-precision robot will be used to control a camera. Also, the exercises will require the implementation and evaluation of one of the decribed methods.
About 15 lectures (each 2 x 45 minutes) will be given Wednesday 10-12, first time February 5 at room N034. The lectures will be given in Danish unless English is demanded. There will be no course fee.
A credit of 7.5 ECTS will be given provided that the two exercises are answered satisfactory.
Formally, it is a prerequisite that a course in Digital image processing has been passed. In practice significant mathematical maturity is needed, e.g. linear algebra will be used extensively. Also basic elements of differential calculus, statistics, differential geometry etc. will be used. Having passed a course in Graphics may be an advantage. I expect that the students will be able to program in C (or similar language). Also, knowledge on MATLAB may show up to be an advantage.
E. Trucco, A. Verri: Introductory Techniques for 3-D Computer Vision
Prentice Hall 1998.
Supplementary material will be handed out when needed.