Understanding Ocean Surface Temperature Features

Spasic, Nemanja and Tilanus, Jared (2005) Understanding Ocean Surface Temperature Features, CS05-07-00, Department of Computer Science, University of Cape Town.

[img] PDF
Technical_Report.pdf

Download (360kB)

Abstract

The aim of this project was to develop a prediction system that uses Artificial Intelligence, machine learning using training data and Image Processing (AI) to extract training data from Sea Surface temperature (SST) images to predict the ocean surface, temperature features around the coast of the Southern African region. Region growing and histographic algorithms were used in the image processing section to extract thermal fronts as training data from the available SST images. A Temporal Bayesian Network was developed as the prediction model which used approximate stochastic learning and inference algorithms based on the Maximum Likelihood Algorithm (MLE). User-Centered Design (UCD) and Human-Computer Interaction (HCI) methods were used to develop user-friendly and easy to understand Graphical User Interfaces (GUI). Results and evaluations of the project revealed that a generally successful prototype implementation of a prediction system that used AI, machine learning and image processing was developed.

Item Type: Technical report
Uncontrolled Keywords: Ocean Surface Temperature Feature Prediction,Image Processing, Region Growing Algorithm, Histographic Algorithm, Bayesian Nwetworks, Temporal Bayesian Networks, Approximate Learning and Inference, Poisson Distribution
Subjects: Human-centered computing
Computing methodologies > Computer graphics > Image manipulation > Image processing
Mathematics of computing > Probability and statistics
Information systems > Data management systems
Computing methodologies > Artificial intelligence
Date Deposited: 21 Oct 2005
Last Modified: 10 Oct 2019 15:35
URI: http://pubs.cs.uct.ac.za/id/eprint/259

Actions (login required)

View Item View Item