Geometric Information Criterion for Model Selection

  • Authors:
  • Kenichi Kanatani

  • Affiliations:
  • Department of Computer Science, Gunma University, Kiryu, Gunma 376 Japan. E-mail: kanatani@cs.gunma-u.ac.jp

  • Venue:
  • International Journal of Computer Vision
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

In building a 3-D model of the environment from image and sensordata, one must fit to the data an appropriate class of models, which can beregarded as a parametrized manifold, or geometric model,defined in the data space. In this paper, we present a statistical frameworkfor detecting degeneracies of a geometric model byevaluating its predictive capability in terms of the expectedresidual and derive the geometric AIC. We showthat it allows us to detect singularities in a structure-from-motionanalysis without introducing any empirically adjustable thresholds. Weillustrate our approach by simulation examples. We also discuss theapplication potential of this theory for a wide range of computer vision androbotics problems.