Date of Award
Master of Applied Science (MASc)
Electrical and Computer Engineering
The multiple hypotheses tracker (MHT) is an optimal tracking method due to the enumeration of all possible measurement-to-track associations. However, its practical implementation is limited by the NP-hard nature of this enumeration. To bound the computational complexity, some means of limiting the number of possible associations is required. Typical solutions include the interposition of rules to guide the pruning and merging of tracks. Other proposals have shown that the performance of a tracker, MHT or not, can be improved using feature information (e.g., signal strength, size, type) in addition to kinematic data. The inclusion of feature information allows for the discrimination to further gate the data associations. However, in most tracking systems, the schemes to manage the data association problem are extraneous to the Bayesian framework of the MHT. Further, the extraction of features from the raw sensor data is typically independent of the subsequent association and filtering stages. The features are then used in either an ad hoc way or are they are fused with the MHT tracker; they are not used intrinsically within MHT framework. In this thesis, a new approach whereby there is an intrinsic interaction between feature extraction and the MHT is presented. The measure of the quality of feature extraction is input into measurement-to-track association while the prediction step feeds back information to be used in the next round of feature extraction to increase the information available a priori. The motivation for this forward and backward interaction between feature extraction and tracking is to improve the performance in both steps. This approach allows for a more rational partitioning of the feature space, removing unlikely features from. the assignment problem. In addition, a track-specific detection probability becomes available to the prior. This probability significantly improves the coasting behavior when measurements are not available for track continuation. Simulation results demonstrate the benefits of the proposed approach.
McAnanama, James, "A Multiple Hypothesis Tracker with Interacting Feature Extraction" (2010). Open Access Dissertations and Theses. Paper 4216.
McMaster University Library