Colloquium - 22.10.2015
Finding geometric structures in images: from human visual perception to computer vision
|Speaker||Prof. Dr. Agnès Desolneux (Centre de Mathématiques et Leurs Applications, Ecole Normale Supérieure, Cachan, France|
|Time||Thursday 22 October 2015 from 17:15 – 18:15 o’clock|
|Location||Department of Mathematics and Computer Science, Seminar Room 00.003, University of Basel, Spiegelgasse 1, Basel|
The a contrario method is a generic method to detect meaningful geometric structures in images. It is based on the combination of the Gestalt Theory of human visual perception and of the so-called non-accidentalness principle (that roughly says that "we immediately perceive in an image what has a low probability of coming from an accidental arrangement''). I will first show in detail how the a contrario method works in the case of the detection of straight segments in an image. Then, I will show some other detection problems. At the end of the talk, I will also explain how the a contrario detection method can, using a maximum entropy principle, be turned into a method able to generate new images having the same visual content as an original given image.
After the talk an informal apéro will be offered. I hope you’ll be joining us.