Open Collections

Using range data for object recognition in a robotics environment

University of British Columbia

A new approach to object recognition for a robotics environment is presented. 2-D models enriched with 3-D information are constructed automatically from a range image. These 'view models' are used to recognize objects by matching them to models subsequently constructed from similar images. A segmentation method for extraction of planar surfaces from range images has been developed. View models are constructed as augmented surface adjacency graphs. A heuristic model matching method, is presented. Results indicate that object recognition in a constrained domain and environment can be achieved at a very high success rate using singlevview range images.

Text

2010-05-09

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/24539

Computer Science

University of British Columbia

Graduate