Adaptive transform for manifold-valued data: The software consists of three main parts: 1) An implementation of the (bandlimited) α-shearlet transform (in AlphaTransform.py, in three versions: A fully sampled (non-decimated), translation invariant, fast, but memory-consuming implementation; A fully sampled, translation invariant, slightly slower, but memory-efficient implementation. A subsampled (decimated), not translation invariant, but fast and memory-efficient implementation. 2) Implementations (in AdaptiveAlpha.py) of three criteria that can be used to adaptively choose the value of α, namely: The asymptotic approximation rate (AAR), the mean approximation error (MAE), the thresholding denoising performance (TDP). 3) A chart-based implementation (in SphereTransform.py) of the α-shearlet transform for functions defined on the sphere.
Keywords for this software
References in zbMATH (referenced in 2 articles )
Showing results 1 to 2 of 2.
- Bubba, Tatiana A.; Kutyniok, Gitta; Lassas, Matti; März, Maximilian; Samek, Wojciech; Siltanen, Samuli; Srinivasan, Vignesh: Learning the invisible: a hybrid deep learning-shearlet framework for limited angle computed tomography (2019)
- Arridge, Simon R. (ed.); de Hoop, Maarten V. (ed.); Maaß, Peter (ed.); Schönlieb, Carola-Bibiane (ed.): Mini-workshop: Deep learning and inverse problems. Abstracts from the mini-workshop held March 4--10, 2018 (2018)