Vision par ordinateur

Objectives

This course covers the notions of calibration, interest point detection (in mono or multi-resolution), matching (global and local) and tracking. In addition, you will learn about the well-known SIFT (Scale Invariant Feature Transform) approach and a classical KLT (Kanade-Lucas-Tomasi) tracking approach.

Description

This part is composed of 2 reversed classroom lessons in order to allow the learner to be more active in his learning. Then, 4 practical works illustrate the notions of detection and matching discussed in class in order to build a mosaic of images. This subject will be evaluated via an online course questionnaire and an exam on paper as well as a grade for the practical work. This allows a continuous evaluation of the acquired knowledge.

Targeted skills

To know the calibration approaches

To know the methods of detection of points of interest and how to use them

To know the different mapping techniques and to know how to handle them

Bibliography

Richard Szeliski. Computer vision: Algorithms and Applications, 2010.

http://szeliski.org/Book/

Pre-requisites

To have followed the second year course Image, Modeling and Rendering or to have notions of image processing and segmentation.