From generalisation to segmentation: Douglas-Peucker-Ramer and movement data
2019-09-16T06:37:41Z (GMT) by
With an explosion in the sensorisation of moving objects (people, cars, wildlife) that has come from the ease of use and acquisition of low-cost GNSS receivers, the volume of tracking data presents challenges for efficient data processing. For example, in analysis and modelling, tracks often need to be segmented based on particular characteristics (e.g., moving vs. stationary). Here we adapt a commonly used polyline generalisation algorithm (Douglas-Peucker-Ramer) to segment tracking data into uphill and downhill portions of the track based on vertex coordinates in terms of their elevation and distance between neighbouring vertices along a track (relative to start of the track). This adaptation supports robust segmentation even for tracks in complex terrain in order to best capture real-world conditions. We present our adapted algorithm and a case study with volunteered backcountry skier tracking data to demonstrate the performance of the algorithm. We conclude with thoughts on future development of tracking data segmentation algorithms.