GIT-GVU-93-07

Image Understanding Research Overview

Daryl T. Lawton

Our work is concerned with several areas of image understanding: perceptual organization, motion processing, model-based vision, image understanding software, and applications in domains such as terrestrial robotics, industrial, and biomedical image processing. This paper is an overview of the different papers from our lab in the 1993 DARPA Image Understanding Workshop Proceedings. We first present several different motion processing algorithms. These algorithms involve generalizing earlier work for processing translational image sequences to less restricted motions, extensions to factorization methods to allow for linear features which are less dependent on precise feature-point matching, and the incorporation of models in processing dynamic images. We then describe the design and initial prototyping of the user interface of the DARPA Image Understanding Environment (IUE) and the tools for documentation, tutorials, and publication that will facilitate the use and adoption of the IUE. We finally present a set of new algorithms for range-free qualitative navigation which will enable mobile robots with limited recognition capabilities to form effective spatial maps for navigation and exploration.

Image understanding, perceptual organization, motion processing, model-based vision, biomedical image processing, Image Understanding Environment (IUE), navigation

PDF

Postscript