The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #92107

Title:Automatic Model Building from Segmented Range Data
Authors:Bispo,EM
Date: 1992
Presented:
Keywords:
Abstract:This dissertation describes the development of a software system for automatic generation of models of man made objects using the Suggestive Modeling System (SMS) created in the IMAGINE project. The dissertation presents a complete algorithm for generating models from range images of view of an object, and also some simple techniques for merging models obtained from distinct views of the same object. The models were built using as input a description of the segmented image that enumerates all the different surfaces present in the object, their boundaries, their position in a common reference frame, their shape (plane, cylinder, etc), their area, perimeter, etc. This information was used to place the object surfaces in the SMS model to obtain simplified approximations of the surface boundaries, to find the intersection between the adjacent surfaces, and to add some SMS properties to the model in order to allow its use for object recognition in the IMAGINE project. The merging of partial models from different views were done by supplying by hand the model feature correspondences between the different views and using this information for calculating the registration between the two views and integrating them. The algorithm successfully generated SMS models of some simple and one complex object that were then used for recognition in the context of the IMAGINE project.
Download:NO ONLINE COPY

[Search These Pages] [DAI Home Page] [Comment]