Wavelet transform
Definition: Wavelet transform (WT) applies an orthogonal kernel to provide a transformation from time space to time-scale space that can be either performed in a discrete or continuous fashion. Note that the term time here is not intended in the physical sense, and can refer to other sampling lattices, for instance, images. By translating and dilatating the kernel, termed the mother wavelet, the wavelet transform provides a time-scale representation of the input data that often facilitates downstream analysis such as spectral analysis or to calculate connectivity estimators. There are many wavelet transforms as well as many mother wavelets with different virtues and assumptions. For instance, the discrete wavelet transform (DWT) performs the dilatation in a less fine-grained resolution as compared to the continuous wavelet transform but it is computationally faster. Wavelets have proved popular in fNIRS for filtering and denoising (e.g. motion artifacts), hyperscanning coherence analysis, or the interpretation of connectivity among others.
Alternative definition:
Synonym:
References: https://doi.org/10.1175/1520-0477(1998)079%3C0061:APGTWA%3E2.0.CO;2
Related terms: Functional Connectivity, Motion Artifact Correction, Time-Frequency Analysis, Preprocessing, Filtering, Multiresolution Analysis