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


Research Paper #582

Title:Experiments in Curvature-Based Segmentation of Range Data
Authors:Trucco,E; Fisher,RB
Date: 1992
Presented:Submitted to the British Machine Vision Conference, Leeds, September 1992
Keywords:
Abstract:Quantitative evaluation methods for computer vision systems are still rather unstructured and subjective. This paper contributes an example of experimental assessment for a range image segmentation system. The tested system is based on estimates of the sign of the mean and gaussian curvatures. The tests reported are aimed at estimating quantitatively the distortions introduced by the segmentation and the overall classification performance of the system. such tests are of primary importance for practical applications and for the choice of thresholds.
Download:POSTSCRIPT COPY


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