TSProcessor

This Time-Series Processor is part of my Bachelors Thesis on Hierarchisches Clustering von Kinect-Bewegungsdaten mittels Dynamic Time Warping (Hierarchical Clustering of Kinect-Motion-Data using Dynamic Time Warping). It can be used to cluster Time-Series Data (e.g. Kinect Data).

Usage

The whole project is written in plain Java. So you just need to install Java. The project has two start parameters.

• The path of the config-file (mandatory).
• A prefix for the paths in the config-file, like "D:" (optional).

Configuration File

The config-File contains all the information necessary to cluster the data. Make sure to avoid the colon sign ":" in this file. Use the second parameter for the prefix instead. The following attributes have to be specified:

• inputPath: The location of the dataset.
• outputPath: The location you want to save the results to.
• separator: The separator used in the dataset to separate values. e.g. "#".
• datasetType: The type of dataset. e.g. "kinect".
• threshold: The threshold for clustering. If the cost is greater than this threshold the clustering will be terminated.
• attributes: The list of attributes this dataset offers.
• usedAttributes: The subset of attributes you want to consider for the cost calculation.
• flipVisualization: This attribute can be used to flip the visualization if the x-values are mirror inverted in this dataset.
• attributeForBodyIdentification: The name of the parameter this dataset uses for the identification of different bodies.
• distanceFunction: The name of the distance function you want to use (You can implement new ones).
• skipFrames: Whether if every third frame should be ignored to increase performance.

An example can be seen in: Example Config-File.

Feedback

If you have any questions or feedback, please reach out to me at laurenz.fuchs@unibw.de.