Nelson, James and Gibberd, Alex and Nafornita, Corina and Kingsbury, Nick (2018) The locally stationary dual-tree complex wavelet model. Statistics and Computing, 28 (6). pp. 1139-1154. ISSN 0960-3174
the_locally_stationary_dual_tree_complex_wavelet_model_nelson_2017.pdf - Published Version
Available under License Creative Commons Attribution.
Download (2MB)
Abstract
We here harmonise two significant contributions to the field of wavelet analysis in the past two decades, namely the locally stationary wavelet process and the family of dual-tree complex wavelets. By combining these two components, we furnish a statistical model that can simultaneously access benefits from these two constructions. On the one hand, our model borrows the debiased spectrum and auto-covariance estimator from the locally stationary wavelet model. On the other hand, the enhanced directional selectivity is obtained from the dual-tree complex wavelets over the regular lattice. The resulting model allows for the description and identification of wavelet fields with significantly more directional fidelity than was previously possible. The corresponding estimation theory is established for the new model, and some stationarity detection experiments illustrate its practicality.