Uncertain Geometry in Computer Vision

Abstract. We give an overview of the main ideas and tools that have been employed in uncertain geometry. We show how several recognition problems in computer vision can be translated into combinatorial optimization problems that involve intersection hypergraphs, and how we can obtain approximate solutions for these problems when we replace the hypergraphs by intersection graphs. The statistical properties of these graphs are important when we design algorithms for the extraction of geometric primitives from images. We illustrate the use of uncertain geometry with examples involving the detection of circles and the computation of transformations between images.

• Type: Conference
• Year: 2005
• Where: DGCI
• Authors: Peter Veelaert
• Keywords: Combinatorial Optimization Problems, DGCI 2005, Intersection Hypergraphs, Uncertain Geometry