Fast Multidimensional Signal Processing using Julia with Shearlab.jl

on YouTube ↗35m

" summary

Shearlab.jl implements the Shearlet Transform and wavelets in Julia and shows concrete code for filtering, FFT use, GPU parallelism and iterative hard thresholding for denoising and inpainting. Benchmarks claim benchmarks vs MATLAB with at least 2× speedups and much larger GPU gains on big images. The library is already used for Light Field Recovery from sparse photographic samples and includes example notebooks for image decomposition, denoising and inpainting and a registered package Shearlab.jl.

AI-written summary · about

" chapters (17)

1. 0:00Julia and Shearlab.jl intro
2. 1:07What is a signal
3. 3:14Fourier transform limits
4. 5:45Short time Fourier vs Gabor
5. 8:17Wavelet transform
6. 9:44Installing Shearlab.jl
7. 14:492D wavelet limits
8. 15:32Cartoon functions and sparsity
9. 18:03Shearlet transform intro
10. 21:54Shearlet implementations vs Julia
11. 23:40Why Julia for signal processing
12. 24:43Implementing Shearlet in Julia
13. 26:54Shearlab.jl vs MATLAB benchmarks
14. 28:20Denoising images with Shearlets
15. 29:24Inpainting with Shearlets
16. 31:143D light field recovery
17. 32:42Q&A

" related talks

• 27mJulia (workshop)Ranjan Anantharaman, Shashi Gowda · Cypher 2017
• 31mBridging Scientific and Industrial Worlds for Advanced Signal and Image ProcessingPierre Raybaut · PyData Paris 2024
• 42mLightFields.jl: Fast 3D image reconstruction for VR applicationsHector Andrade Loarca · PyData Berlin 2018
• 01h 38mAn Introduction to JuliaJane Herriman · PyData Los Angeles 2018
• 43mJulia for Data Analysis features, interfaces and future directionsSimon Byrne · PyData London 2016
• 32mJulia for PythonistasKelly Shen · PyData New York City 2019
• 47mThe many ways of parallel computing with JuliaViral B Shah · The Fifth Elephant 2015
• 43mJulia: A Fresh Approach to Machine LearningMike Innes · PyData London 2017