"Clustering is an important task in applications involving classification of physical or abstract objects into different classes or groups. Many clustering algorithms from different approaches like machine learning, statistics and databases are available to perform clustering.
For the landmine detection and discrimination problem, the objects present in the ground are studied with different sensors. The noisy outputs in the form of signals or images from these sensors are preprocessed for clustering. The challenge is to cluster the data without having sufficient information about the data.
In this thesis, methods for preprocessing the data and a dynamic unsupervised clustering algorithm (DUCA) are proposed. This algorithm is adaptive, dynamic in nature with dynamic creation and merging of clusters, and the evolving nature of the clusters would result in widths that cover the real clusters and classifies the future patterns correctly.
A performance analysis on the simulated data is conducted. Our study shows that the algorithm could identify clusters properly and considered outliers as noise. This algorithm is efficient and effective. Additionally it is compared with the other unsupervised clustering algorithms"--Abstract, page iii.
Rao, Vittal S.
McMillin, Bruce M.
M.S. in Computer Science
University of Missouri--Rolla
x, 99 pages
© 2000 Mereddy Pramodh Kumar Reddy, All rights reserved.
Thesis - Restricted Access
Print OCLC #
Electronic OCLC #
Link to Catalog Record
Electronic access to the full-text of this document is restricted to Missouri S&T users. Otherwise, request this publication directly from Missouri S&T Library or contact your local library.http://merlin.lib.umsystem.edu/record=b4446301~S5
Reddy, Mereddy Pramodh Kumar, "Dynamic unsupervised clustering algorithm (DUCA) for sensor data fusion" (2000). Masters Theses. 1939.
Share My Thesis If you are the author of this work and would like to grant permission to make it openly accessible to all, please click the button above.