Evaluating distance measures for trajectories in the mobile setting

Postgraduate Thesis uoadl:1321050 528 Read counter

Unit:
ΠΜΣ Πληροφορικής και Τηλεπικοινωνιών με ειδίκευση Προηγμένα Πληροφοριακά Συστήματα
Library of the School of Science
Deposit date:
2016-07-18
Year:
2016
Author:
Μητατάκης Χρήστος
Supervisors info:
Δημήτριος Γουνόπουλος
Original Title:
Evaluating distance measures for trajectories in the mobile setting
Languages:
English
Translated title:
Αξιολόγηση μέτρων απόστασης για σύγκριση τροχιών μέσω κινητών συσκευών
Summary:
The study of the similarity between trajectories is important in a plethora of
application domains (e.g. Traffic management, Video analysisetc.). In this
thesis a new approach of Frechet Distance algorithm was implemented. We
initiate an experimental evaluation of three different trajectory similarity
measures, namely Dynamic Time Warping (DTW), Longest Common Subsequence (LCSS),
and Frechet distance. Intuitively, Dynamic/ Time Warping (DTW) is the
equivalent of the L2 distance when stretching of the trajectories is allowed
since DTW takes into account the individual differences of all matched points
in the stretched sequences. Similarly, Longest Common SubSequence (LCSS) can be
thought of as the equivalent of L0 since it counts how many elements are the
same. Completing the analogy, the Frechet distance is equivalent/ to Linf since
it considers the maximum of the individual differences of the matches in the
stretched sequences.
Keywords:
Euclidean Distance, Frechet Distance, Dynamic Time Warping, Longest Common Subsequence, Row Data
Index:
Yes
Number of index pages:
8-11
Contains images:
Yes
Number of references:
14
Number of pages:
68
File:
File access is restricted.

document.pdf
6 MB
File access is restricted.