Roach, J. W. "Determining the Three-Dimensional Motion and Model of Objects from a Sequence of Images." The University of Texas at Austin, Department of Computer Sciences. AI Technical Report 80-1. October 1980.
The goal of this dissertation is to determine precisely how an object is moving in three-dimensional space and to determine the three-dimensional- relationship of points on the surface of the object. The only information available is a sequence of photographic images taken as the object moves across the field of view. The problem can be broken down into two sub-problems: the problem of determining the correspondence of feature points in one image with feature points in the next image; and once the correspondence is established, the mathematical analysis required to determine the model and the movement. The correspondence problem, i.e., matching, is investigated using images of moving blocks. The corners of the blocks constitute the feature points to be put in correspondence between images. Several matching methods are combined into a hierarchy so that if one method fails another method can take over to help complete the matching process. The top level of the hierarchy matches by searching for feature points in the image of expected positions as computed from the expected movement of the object. The next hierarchy level matches an object by its position relative to other objects in the image, a property that is assumed to change only gradually. The next hierarchy level matches a block's faces by relative position. Once faces have been matched, feature points bordering the faces not already matched by expected position can be put in correspondence. The mathematical analysis of the problem shows that there are an infinite number of geometrically similar solutions, each solution differing from the others by a scaling factor. A specific solution can be found by setting the scaling to an arbitrary number. Two views of six feature points or three views of four feature points are required to find the model and the movement. For good accuracy, however, considerably more points, two views of twelve or fifteen points for example, are needed. Also available as TR80-02.
Martin, W. N. and J. K. Aggarwal. "Occlusion in Dynamic Scene Analysis." The University of Texas at Austin, Department of Computer Sciences. AI Technical Report 80-1b. September 1980. 14 pages.
Originally issued as ICSCA-CMP-2. Abstract not available.
Martin, W. N., J. K. Aggarwal, and Donald I. Good. "A Preliminary Evaluation of Verifiability in ADA." The University of Texas at Austin, Department of Computer Sciences. AI Technical Report 80-18. November 1980. 19 pages.
Also known as report number: ICSCA-CMP-18
Young, William D. and Donald I. Good. "Generics and Verification in ADA." The University of Texas at Austin, Department of Computer Sciences. AI Technical Report 80-19. December 1980. 16 pages.
Originally issued as ICSCA-CMP-19. Abstract not available.
Smith, Lawrence Mark. "Compiling from the Gypsy Verification Environment." The University of Texas at Austin, Department of Computer Sciences. AI Technical Report 80-20. August 1980. 75 pages.
Originally issued as ICSCA-CMP-20. Abstract not available.
Questions to trcenter@cs.utexas.edu