Date

2022-2-10

Author

Tuncer, Barkın

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

363
views

216
downloads

Cite This

In this thesis, we investigate two major topics; Random matrix-based models using variational Bayes inference for extended target tracking and shape classification algorithms for tracking applications. In the scope of this thesis, we have derived two novel random matrix-based tracking algorithms. First, to represent the extent of dynamic objects as an ellipsoid with a time-varying orientation angle, and secondly, to estimate the extent of the object or a group of objects with more than one ellipsoid. In both of these solutions, we have used the variational Bayes technique to perform approximate inference, where the Kullback-Leibler divergence between the true and the approximate posterior is minimized by performing fixed-point iterations. The update equations are easy to implement, and the algorithms can be used in real-time tracking applications. We illustrated the performance of the methods in simulations and experiments with real data. In extended target tracking, once the shape estimate of an object is formed, it can naturally be utilized by high-level tasks such as classification of the object type. Therefore, we present a naively deep neural network, which consists of one input, two hidden and one output layers, to classify dynamic objects regarding their shape estimates. In this manner, the proposed method shows superior performance in comparison to a Bayesian classifier for simulation experiments.

Subject Keywords

Extended target tracking, Object tracking, Variational Bayes, Kalman filter, Machine learning, Classification, Target classification, Neural networks

URI

https://hdl.handle.net/11511/96370

Collections

Graduate School of Natural and Applied Sciences, Thesis