Type of Document Master's Thesis Author Paripati, Praveen Kumar URN etd-06102012-040611 Title Polyhedra:representation and recognition Degree Master of Science Department Computer Science and Applications Advisory Committee
Advisor Name Title Roach, John W. Committee Chair Bixler, J. Patrick Committee Member Ehrich, Roger W. Committee Member Heath, Lenwood S. Committee Member Keywords
Date of Defense 1989-05-15 Availability restricted AbstractComputer Aided Design systems intended for three dimensional solid modelling
have traditionally used geometric representations incompatible with established representations
in computer vision. The utilization of object models built using these
systems require a representation conversion before they can be used in automatic
sensing systems. Considerable advantages follow from building a combined CAD
and sensing system based on a common geometric model. For example, a library
of objects can be built up and its models used in vision and touch sensing system
integrated into an automated assembly line to 'discriminate between objects and determine-
orientation and distance. This thesis studies a representation scheme, the
dual spherical representation, useful in geometric modelling and machine recognition.
We prove that the representation uniquely represents genus 0 polyhedra. We
show by,example that our representation is not a strict dual of the vertex connectivity graph, and hence is not necessarily ambiguous. However, we have not been able to prove that the representation is unambiguous. An augmented dual spherical
representation which is unique for general polyhedra is presented. This graph theoretic
approach to polyhedra also results in an elegant method for decomposition of polyhedra into combinatorially convex parts. An algorithm implementation details and experimental results for recognition of polyhedra using a large field tactile
sensor are given. A theorem relating the edges in the dual spherical representation and the edge under perspective projection is proved. Sensor fusion using visual and tactile sensory inputs is proposed to improve recognition rates.
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access LD5655.V855_1989.P374.pdf 5.92 Mb 00:27:25 00:14:06 00:12:20 00:06:10 00:00:31next to an author's name indicates that all files or directories associated with their ETD are accessible from the Virginia Tech campus network only.
If you have questions or technical problems, please Contact DLA.