Mining continuous activity patterns from animal trajectory data

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
By: , and 

Links

Abstract

The increasing availability of animal tracking data brings us opportunities and challenges to intuitively understand the mechanisms of animal activities. In this paper, we aim to discover animal movement patterns from animal trajectory data. In particular, we propose a notion of continuous activity pattern as the concise representation of underlying similar spatio-temporal movements, and develop an extension and refinement framework to discover the patterns. We first preprocess the trajectories into significant semantic locations with time property. Then, we apply a projection-based approach to generate candidate patterns and refine them to generate true patterns. A sequence graph structure and a simple and effective processing strategy is further developed to reduce the computational overhead. The proposed approaches are extensively validated on both real GPS datasets and large synthetic datasets.

Additional publication details

Publication type Article
Publication Subtype Journal Article
Title Mining continuous activity patterns from animal trajectory data
Series title Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
DOI 10.1007/978-3-319-14717-8_19
Volume 8933
Year Published 2014
Language English
Publisher Springer
Publisher location New York, NY
Contributing office(s) Patuxent Wildlife Research Center
Description 14 p.
First page 239
Last page 252
Online Only (Y/N) N
Additional Online Files (Y/N) N