Michael Brady

Shape from symmetry: detecting and exploiting symmetry in affine images

We investigate the constraints placed on the image projection of a planar object having local reflectional symmetry. Under the affine approximation to projection, we demonstrate an efficient (low-complexity) algorithm for detecting and verifying symmetries despite the distorting effects of image skewing. The symmetries are utilized for three distinct tasks: first, determining image back-projection up to a similarity transformation ambiguity; second, determining the object plane orientation (slant and tilt); and third, as a test for non-coplanarity amongst a collection of objects. These results are illustrated throughout with examples from images of real scenes.

Stereo matching of curves

Abstract A stereo algorithm which matches connected chains of edgels (curves) between images is described. It is based on representing the curves as elastic strings/snakes, and measuring the amount of deformation the strings have to undergo to transform between corresponding curves, and incorporates the ideas of the disparity gradient, and the fact that matching sections of curve have to be of a similar shape. This explicit use of shape information means that a precisely known epipolar geometry is no longer crucial. Pairs of potentially corresponding curves which lead to a large deformation energy, are eliminated, and the greatly reduced number of potentially matching pairs are passed on to a tree search stage. A typical result of running the algorithm on a stereo triple is presented.

Practical Structure and Motion from Stereo When Motion is Unconstrained

This paper describes a system which robustly estimates motion, and the 3D structure of a rigid environment, as a stereo vision platform moves through it. The system can cope with any camera motion, and any scene structure and is successful even in the presence of large jumps in camera position between the capture of successive image pairs, and when point matching is ambiguous. The system was developed to provide robust obstacle avoidance for a partially sighted person.The process described attempts to maximise use of the abundant information present in a stereo sequence. Key features include the use of multiple stereo match hypotheses, efficient motion computation from three images, and the use of this motion to ensure reliable matching, and to eliminate multiple stereo matches. Points are reconstructed in 3D space and tracked in a static coordinate frame with a Kalman Filter.This results in good 3D scene reconstructions. Structure which is impossible to match with certainty is absent, rather than being incorrectly reconstructed. As a result, the system is appropriate for obstacle detection. The results of processing some indoor and outdoor scenes, are given in the paper, and practical issues are highlighted throughout.

Seeds of Perception.

Not all information is created equal. Different locations in an image (or range image) impose differing levels of constraint upon visual processes. We call locations in images (and other representations) that offer the tightest constraint seeds. We are exploring algorithms that work initially from seeds to locations of ever decreasing constraint. Thus in motion, we work from corners to edges and then to regions bounded by those edges. We recount recent progress in developing this theme in the problems of computing shape from contour, the determination of optic flow and structure from motion, and stereo. We review related work in shape from shading, and the interpretation of variable geometry planar shapes. The loci of two dimensional change that are discussed in this paper are uncovered by the programs of Fleck and Noble at this conference, as well as by Forsyths colour labelling of edges.

A parallel implementation of a structure-from-motion algorithm

This paper describes the implementation of a 3D vision algorithm, Droid, on the Oxford parallel vision architecture, PARADOX, and the results of experiments to gauge the algorithms effectiveness in providing navigation data for an autonomous guided vehicle. The algorithm reconstructs 3D structure by analysing image sequences obtained from a moving camera. In this application, the architecture delivers a performance of greater than 1 frame per second — 17 times the performance of a Sun-4 alone.

Stereo Matching of Curves.

This paper describes a stereo algorithm which matches connected chains of edgels (curves) between images. It is based on representing the curves as elastic strings and measuring the amount of deformation the strings have to undergo to transform between corresponding curves, and incorporates the ideas of the disparity gradient and the fact that matching sections of curve have to be of a similar shape. This explicit use of shape information means that a precisely known epipolar geometry is no longer crucial. Pairs of potentially corresponding curves which lead to a large deformation energy, are eliminated and the greatly reduced number of potentially matching pairs are passed on to a tree search stage. A typical result of running the algorithm on a stereo triple is presented.