Abstract
In this work, it is demonstrated that covariance estimator methods can be used for trajectory classification. It is shown that, features obtained via shrunk covariance estimation are suitable for describing trajectories. Compared to Dynamic Time Warping, application of explained technique is faster and yields more accurate results. An improvement of Dynamic Time Warping based on counting statistical comparison of base distance measures is also achieved. Results on Australian Sign Language and Character Trajectories datasets are reported. Experiment realizations imply feasibility through covariance attributes on time series.DOI: http://dx.doi.org/10.5755/j01.eie.24.3.15290
Highlights
In this work, focus is directed on the usage of covariance matrices for trajectory categorization
Basic covariance features (which were induced from the initial form of (11), where we had no parameter at all) tested on Australian Sign Language (ASL) collection [14]
Trajectory analysis is beneficial for many applications ranging from driver intention forecasting [9] to aircraft engineering [19]
Summary
Focus is directed on the usage of covariance matrices for trajectory categorization. No shrunk covariance technique is applied to time series, especially in the categorization task. From Hidden Markov Models [5] and ‘elbow’ reductions [6] to kernel-based representations [7], there are various ways of handling trajectory analysis tasks but none of these are covariance-based except [8], where covariance matrices are instrumented to develop DTW rather than extracting explicit vector set descriptions. In terms of feature formation, that is, deduction of fixed-dimensional vectors from ordered vector sets, literature is still lacking in covariance-based analysis; especially when it comes to shrunk models. For we have tasks of manoeuvre [9] and driver intention [10] forecasting Such kind of security applications need fast spatial time-series categorization. In the third section, an analysis with potential future work route suggestion is demonstrated
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.