Directional wavelet packets originating from polynomial splines (Q6042631)

The paper presents a collection of quasi-analytic wavelet packets and discusses their properties, in particular with respect to their application in image processing (e.g., for denoising, inpainting, and classification tasks). An important aspect in this context is how these wavelet packets behave under the operation of a discrete Fourier transform. The authors argue that their concepts are also useful in connection with deep learning techniques.