Recovering Free Space from a Single Two-Point Perspective Image for Mobile Robot Navigation for Indoor Applications

The problem of identifying free space in an environment from a given single image is a subject matter of interest in the field of computer vision and robotics. The aim of this paper is to construct a scaled 2D floor map from a single two-point perspective (2PP) image for visualizing the relative dimensions and poses of the objects, by taking minimal information. This is done by proposing a new simple method which utilizes the perspective property of the given image, viz., vanishing points. This paper introduces necessary geometrical constructions by introducing the notion of the side view of an image, using which few simple geometrical relations will be derived, that will help in constructing a 2D floor map. To the best of our knowledge, this is the first work of its kind, which introduces architectural concepts for map construction and for measurements from it, thus proving novelty of the introduced method.