Data Mining On Animal Moving Trajectories

Researches on animal moving trajectory data mining and feature extraction have played a more and more important role in biodiversity protection, environmental conservation, disease prevention and many other aspects. At the same time, with the fast development and wild application of wireless tracking, satellite communication and GPS satellite positioning technology, there comes more and more types of animal moving trajectory, and the amount of trajectory data has become huger and huger Therefore, it has a great importance to choose a proper way to analyze animal moving trajectory data. This paper gives priority to the feature extraction of animal trajectory stop and trajectory similarity analysis based on the Starkey project data, proposes an improved method based on DBSCAN to extract trajectory stop points and stop areas, and analyzed the moving characteristic of animals. Specifically, main researches of this paper are as the following:1. Spatial hot spots analysis based on gridsThis paper overlay the trajectory data with the grids based on the meaning of the hot spot of animal trajectories, and analyze the stay time of the trajectory data. Then extract spatial hot spots of the Starkey project data and obtained the spatial distribution of these three kinds of animals from 1993 to 1996.2. Extracting trajectory stops based on spatial scope and duration thresholdGenerally speaking, comparing moving speed with a certain threshold is the most popular way to judge whether it is a stop point or not. However, this kind of solution will lead to a severe calculation error on incomplete data. Therefore, this paper gives an improved way based on DBSCAN, which uses two parameters, spatial range and duration, to extract trajectory stops and to do some analysis. Experiments have been done with the improved way on 1993 deer data, and the result showed that this kind of solution could solve the problem very well.3. Trajectory similarity analysisHow to choose a proper measure method of similarity is the priority of trajectory similarity analysis. This paper choose OWD to measure it according to the research data of the actual situation, and hierarchical clustering was applied to Starkey experiment trajectory data after similarity measurement. Trajectories have been classified to several groups after the experiment, and it could be recognized easily.