Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Signal Processing Course: Denoising
Search
Gabriel Peyré
January 01, 2012
Research
0
1.4k
Signal Processing Course: Denoising
Gabriel Peyré
January 01, 2012
Tweet
Share
More Decks by Gabriel Peyré
See All by Gabriel Peyré
The mathematics of transformers
gpeyre
0
150
Computational OT #4 - Gradient flow and diffusion models
gpeyre
0
860
Computational OT #1 - Monge and Kantorovitch
gpeyre
0
960
Computational OT #2 - Entropic regularization
gpeyre
0
460
Computational OT #3 - Dual and semidiscrete
gpeyre
0
300
Optimal and Diffusion Transports in Machine Learning
gpeyre
0
2.3k
The Mathematics of AI
gpeyre
1
2.4k
Transformers are Universal in Context Learners
gpeyre
0
990
Ground Metric Learning with applications in genomics
gpeyre
0
780
Other Decks in Research
See All in Research
ローテーション別のサイドアウト戦略 ~なぜあのローテは回らないのか?~
vball_panda
0
310
Thirty Years of Progress in Speech Synthesis: A Personal Perspective on the Past, Present, and Future
ktokuda
0
190
HU Berlin: Industrial-Strength Natural Language Processing with spaCy and Prodigy
inesmontani
PRO
0
280
20年前に50代だった人たちの今
hysmrk
0
170
台湾モデルに学ぶ詐欺広告対策:市民参加の必要性
dd2030
0
290
Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal Reasoning
satai
3
680
それ、チームの改善になってますか?ー「チームとは?」から始めた組織の実験ー
hirakawa51
0
990
姫路市 -都市OSの「再実装」-
hopin
0
1.7k
離散凸解析に基づく予測付き離散最適化手法 (IBIS '25)
taihei_oki
1
730
[SITA2025 Workshop] 空中計算による高速・低遅延な分散回帰分析
k_sato
0
130
ForestCast: Forecasting Deforestation Risk at Scale with Deep Learning
satai
3
580
空間音響処理における物理法則に基づく機械学習
skoyamalab
0
250
Featured
See All Featured
Avoiding the “Bad Training, Faster” Trap in the Age of AI
tmiket
0
110
XXLCSS - How to scale CSS and keep your sanity
sugarenia
249
1.3M
Max Prin - Stacking Signals: How International SEO Comes Together (And Falls Apart)
techseoconnect
PRO
0
130
<Decoding/> the Language of Devs - We Love SEO 2024
nikkihalliwell
1
160
Design of three-dimensional binary manipulators for pick-and-place task avoiding obstacles (IECON2024)
konakalab
0
390
Lightning talk: Run Django tests with GitHub Actions
sabderemane
0
150
Google's AI Overviews - The New Search
badams
0
950
DevOps and Value Stream Thinking: Enabling flow, efficiency and business value
helenjbeal
1
150
JAMstack: Web Apps at Ludicrous Speed - All Things Open 2022
reverentgeek
1
400
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
47
8k
Put a Button on it: Removing Barriers to Going Fast.
kastner
60
4.2k
Rebuilding a faster, lazier Slack
samanthasiow
85
9.4k
Transcript
Linear and Non Linear Denoising Gabriel Peyré www.numerical-tours.com
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Noise in Images
Denoising Problem
Denoising Problem
Additive Noise Model
Noise Distributions −0.3 −0.2 −0.1 0 0.1 0.2 0.3 −0.3
−0.2 −0.1 0 0.1 0.2 0.3
Noise Distributions −0.3 −0.2 −0.1 0 0.1 0.2 0.3 −0.3
−0.2 −0.1 0 0.1 0.2 0.3
Noise Distributions −0.3 −0.2 −0.1 0 0.1 0.2 0.3 −0.3
−0.2 −0.1 0 0.1 0.2 0.3
Data-dependent Noise
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Linear Denoising Estimator
Fourier and Denoising
Optimal Filter Choice
Oracle Estimation of Optimal Filter
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Diagonal Thresholding
Wavelet Diagonal Hard Thresholding
Sparse Signal Estimation
Optimal Threshold Selection
W unit variance white noise. Non-linear Approximation and Estimation
Hard vs. Soft Thresholding
Hard vs. Soft Thresholding
Optimal Threshold
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Translation Invariant Denoising
Translation Invariant Wavelets
Translation Invariant Haar (1D)
Translation Invariant Transform (2D)
Translation Invariant Thresholding
Optimal Invariant Threshold
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Between Hard and Soft Thresholding
Stein Quadratic-Soft Thresholder
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Block Thresholding
Optimal Block Choice
Comparison
Overview • Noise in Signals and Images • Linear Denoising
by Blurring • Non-linear Wavelet Denoising • Translation Invariant Thresholding • Other Diagonal Thresholders • Non-diagonal Block Thresholding • Data-dependent Noise
Poisson Noise
Poisson Noise Variance Stabilization 1 2 3 4 5 6
7 8 9 10 0.75 0.8 0.85 0.9 0.95 1 1.05
Multiplicative Noise
Multiplicative Noise Stabilization 0 0.5 1 1.5 2 2.5 −1.5
−1 −0.5 0 0.5 1 1.5
Conclusion