Skip to main content

Automated interpretation of digital images of hydrographic charts.

Goodson, K. J., 1987. Automated interpretation of digital images of hydrographic charts. Doctoral Thesis (Doctoral). Dorset Institute of Higher Education.

Full text available as:

[img]
Preview
PDF (pdf supplied by EThOS.)
Goodson,Kelvin_Ph.D._1987.pdf

20MB

Abstract

Details of research into the automated generation of a digital database of hydrographic charts is presented. Low level processing of digital images of hydrographic charts provides image line feature segments which serve as input to a semi-automated feature extraction system, (SAFE). This system is able to perform a great deal of the building of chart features from the image segments simply on the basis of proximity of the segments. The system solicits user interaction when ambiguities arise. IThe creation of an intelligent knowledge based system (IKBS) implemented in the form of a backward chained production rule based system, which cooperates with the SAFE system, is described. The 1KBS attempts to resolve ambiguities using domain knowledge coded in the form of production rules. The two systems communicate by the passing of goals from SAFE to the IKBS and the return of a certainty factor by the IKBS for each goal submitted. The SAFE system can make additional feature building decisions on the basis of collected sets of certainty factors, thus reducing the need for user interaction. This thesis establishes that the cooperating IKBS approach to image interpretation offers an effective route to automated image understanding.

Item Type:Thesis (Doctoral)
Additional Information:A thesis submitted in canditure for the degree of Doctor of Philosophy at the Dorset Institute of Education in collaboration with Marinex Marine Holdings Ltd. If you feel that this work infringes your copyright please contact the BURO Manager.
Group:Faculty of Science & Technology
ID Code:382
Deposited By:INVALID USER
Deposited On:07 Nov 2006
Last Modified:09 Aug 2022 16:02

Downloads

Downloads per month over past year

More statistics for this item...
Repository Staff Only -