Image Denoising and Enhancement -...
Transcript of Image Denoising and Enhancement -...
Department of Signal Processing
1Image Denoising and EnhancementKaren Egiazarian (TUT, NI)
Department of Signal Processing
2
Image denoising: motivating example
• Images are inevitably corrupted by various degradations andparticularly by noise.• Megapixels race: Pixels are getting smaller, and images even noisier
=_
image noise denoised image
Canon Powershot A590IS ISO 800
Department of Signal Processing
3
Imaging Sensors: Exposure-time/noise trade-off
Digital imaging sensors can have very different performance
Different acquisition settings result in different noise levels in the image
“Exposure-time/noise trade-off “
Department of Signal Processing
• Intro• Signal-dependent noise modeling and removal for digital imaging
sensors• Local polynomial approximations (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
4
Outline
Department of Signal Processing
5
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
6
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
7
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations equipped with ICI rule (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
8
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations equipped with ICI rule (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
9
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
10
Outline
• Signal-dependent noise modeling and removal for digital imagingsensors
• Local polynomial approximations (LPA-ICI)• Advanced image processing techniques:
- shape-adaptive methods- nonlocal transform-based methods
• Applications:- denoising- deblurring- deblocking- super-resolution/zooming
Department of Signal Processing
11
Intro
Department of Signal Processing
12
Intro
Department of Signal Processing
13
Intro
Department of Signal Processing
14
Intro
Department of Signal Processing
15
Intro
Department of Signal Processing
16
Karen Egiazarian
Intro
Department of Signal Processing
17
Karen Egiazarian
Intro
Department of Signal Processing
18
Karen Egiazarian
Intro
Department of Signal Processing
19
Karen Egiazarian
Intro
Department of Signal Processing
20
Karen Egiazarian
Intro
Department of Signal Processing
21
Karen Egiazarian
Intro
Department of Signal Processing
22
Karen Egiazarian
Intro
Department of Signal Processing
23
Statistical analysis of raw data
Department of Signal Processing
24
8.4.2016
Statistical analysis of raw data
Department of Signal Processing
25
Statistical analysis of raw data
Department of Signal Processing
26
Statistical analysis of raw data
Department of Signal Processing
27
Statistical analysis of raw data
Department of Signal Processing
28
Statistical analysis of raw data
Department of Signal Processing
29
Statistical analysis of raw data
Department of Signal Processing
30
The analysis of experimental datademonstrates that:
1. The model of noise is close to thePoissonian one
2. Model parameters depend neither on thecolor channel nor on the exposure time
Statistical analysis of raw data
Department of Signal Processing
31
Parametric signal-dependent noise-modelling:Poissonian-Gaussian with clipping
Department of Signal Processing
32Parametric signal-dependent noise-modelling:automatic estimation from single-image raw-data (http://www.cs.tut.fi/~foi/sensornoise.html)
Department of Signal Processing
33
Practical modeling for raw data: idea
• Model photon-to-electron conversion using Poisson distributions (signaldependent);• Model the other noise sources as signal-independent and Gaussian (central-limit theorem);• Exploit normal approximation of Poisson distributions;• The acquisition/dynamic range is limited: too dark or too bright signals areclipped;• There can be a pedestal;• Spatial dependencies can be ignored for normal operating conditions (go forindependent noise).
Eventually, only two parameters are sufficient to describe the noise modelwhere the raw data is described as clipped signal-dependent observations.
Department of Signal Processing
Variance stabilization
34
Karen Egiazarian
Department of Signal Processing
35
Karen Egiazarian
Variance stabilization
Department of Signal Processing
Inversion for Poisson stabilized by AnscombeMäkitalo, Foi (TIP, 2011)
36
Karen Egiazarian
Department of Signal Processing
Experiment: clipped noisy data
37
Original image : y (x1, x2) = 0.7 sin (2π x1/512)+ 0.5
Department of Signal Processing
Experiment: Noise Estimation
38
estimation and fitting a = 0.0038, b = 0.022 st.dev.-function .σ
Department of Signal Processing
Experiment: denoised estimate after variancestabilization before declipping
39
Department of Signal Processing
Experiment: declipped estimate
40
Department of Signal Processing
Experiment: declipped estimate (crosssection)
41
Department of Signal Processing
Real experiment: (Raw-data from Fujiflm FinePixS9600, ISO 1600)
42
Department of Signal Processing
Real experiment: Denoising before declipping
43
Department of Signal Processing
Real experiment: Denoising after declipping
44
Department of Signal Processing
Real experiment: Denoising after declipping(crossection)
45
Department of Signal Processing
46
LASIPwww.cs.tut.fi/~lasip/
• Local Approximation Signal and Image Processing(LASIP) Project
LASIP project is dedicated to investigations in a wide class ofnovel efficient adaptive signal processing techniques.
Department of Signal Processing
47
LASIP
LPA estimates, bias and variance, and asymptotic MSE
Department of Signal Processing
48
LASIP: Intersection of Confidence Intervals (ICI) ruleGoldenshluger & Nemirovski, 1997
Department of Signal Processing
49
Karen Egiazarian
Department of Signal Processing
Anisotropy: motivation
50
Karen Egiazarian
Department of Signal Processing
Anisotropic estimator based on directionaladaptive-scale: idea
51
Karen Egiazarian
Department of Signal Processing
Directional LPA
52
Karen Egiazarian
Department of Signal Processing
53
LASIP: HOW LPA-ICI WORKS
Department of Signal Processing
Anisotropic LPA-ICI:Kernels used in practice
54
Karen Egiazarian
Department of Signal Processing
55
Karen Egiazarian
Department of Signal Processing
56
Karen Egiazarian
Department of Signal Processing
57
Karen Egiazarian
Department of Signal Processing
xiB
yiB
DCT
DCT-1
xiB
yiB
KxK
xiB
yiB
Sliding DCT denoising
K. Egiazarian, J. Astola, M. Helsingius, and P.Kuosmanen (1999) “Adaptive denoising andlossy compression of images in transformdomain”, J. Electronic Imaging
Department of Signal Processing
59
Shape-adaptive DCT image filtering
By demanding the local fit of a polynomial model, we are able to avoidthe presence of singularities or discontinuities within the transform support. In thisway, we ensure that data are represented sparsely in the transform domain,significantly improving the effectiveness of shrinkage (e.g., thresholding).
noisy image and noisy data after hard-thresholdingadaptive-shape extracted from in SA-DCT domainneighborhood the neighborhood
Department of Signal Processing
60
Shape-adaptation: use directional LPA-ICI
Department of Signal Processing
61
Shape-adaptive DCT image filtering
Pointwise SA-DCT: anisotropic neighborhoods
Department of Signal Processing
62
Shape-adaptive DCT image filtering
•Direct generalization of the classical block-DCT (B-DCT);•On rectangular domains (e.g., squares) the SA-DCT and B-DCT coincide;•Comparable computational complexity as the separable B-DCT (fastalgorithms);•SA-DCT is part of the MPEG-4 standard;•Efficient (low-power) hardware implementations available.
Before our work on SA-DCT filtering, the SA-DCT had been usedonly for image and video compression.
Department of Signal Processing
63
Pointwise SA-DCT: denoising results
A fragment of Cameraman: noisy observation (σ=25, PSNR=20.14dB), BLS-GSM estimate (Portilla et al.) (PSNR=28.35dB), and the proposed PointwiseSA-DCT estimate (PSNR=29.11dB).
Department of Signal Processing
64
Pointwise SA-DCT: deblocking results
JPEG coded Cameraman with 2 different quality levels and the results ofpost-filtering using SA-DCT
Department of Signal Processing
65
Pointwise SA-DCT: deblurring results
Images blurred & noisy are deblurred & denoised by SA-DCT filter.
Department of Signal Processing
66
Pointwise SA-DCT: extension to color, motivation
Luminance-chrominance decompositions: structural correlation
Department of Signal Processing
67
Pointwise SA-DCT: structural contraint inluminance-chrominance space
Use for all three channels the adaptive neighborhoods defined by the anisotropicLPA-ICI for the luminance channel.
Department of Signal Processing
68
Pointwise SA-DCT: deblocking results
JPEG-compressed Pointwise SA-DCT deblocking(Q=10, 0.25bpp, PSNR=26.87dB) (PSNR=28.30dB)
Department of Signal Processing
69
Pointwise SA-DCT: deblocking results
Department of Signal Processing
70
Pointwise SA-DCT: denoising results
Fragments of the noisy F-16 (σ=30, PSNR=18.59dB), of ProbShrink-MB(Pizurica et al.) estimate (PSNR=30.50dB), and of Pointwise SA-DCTestimate (PSNR=31.59dB).
Department of Signal Processing
71
Block-Matching and 3D filtering (BM3D)denoising algorithm
• Generalizes NL-means and overcomplete transform methods• Current state-of-the-art denoising method
K. Dabov, A. Foi, V. Katkovnik, and K. Egiazarian, “Imagedenoising with block-matching and 3D filtering”, Proc. SPIEElectronic Imaging 2006, Image Process.: Algorithms and SystemsV, no. 6064A-30, San Jose (CA), USA, Jan. 2006.
--- , “Image denoising by sparse 3D transform-domain collaborativefiltering”, IEEE Trans. Image Process., vol. 16, no. 8, pp. 2080-2095, Aug. 2007.
Department of Signal Processing
72
Block-matching and grouping
Groups are characterized by both:• intra-block correlation between the pixels of each grouped block (naturalimages);• inter -block correlation between the corresponding pixels of different blocks(grouped block are similar);
Department of Signal Processing
73
BM3D: Collaborative filtering
• Each grouped block collaborates for the filtering of all others, and vice versa.• Provides individual estimates for all grouped blocks (not necessarily equal).• Realized as shrinkage in a 3-D transform domain.
Department of Signal Processing
74
BM3D with Shape-Adaptive PCA (BM3D-SAPCA)
Main ingredients:
• Local Polynomial Approximation - Intersection of Confidence Intervals (LPA-ICI) to adaptively select support for 2-D transform;• Block-Matching to enable non-locality;• Shape-Adaptive PCA (SA-PCA);• Shape-Adaptive DCT low-complexity 2-D transform on arbitrarily-shapeddomains (when SA-PCA is not feasible).
K. Dabov, A. Foi, V. Katkovnik, and K. Egiazarian, .BM3D Image Denoisingwith Shape-Adaptive Principal Component Analysis., Proc. Workshop on SignalProcessing with Adaptive Sparse Structured Representations (SPARS.09), Saint-Malo, France, April 2009.
Department of Signal Processing
75
BM3D-SAPCA
Department of Signal Processing
76
Comparison of BM3D-SAPCA with otherfilters
Department of Signal Processing
77
Comparison of BM3D-SAPCA with otherfilters
Department of Signal Processing
78
Comparison of BM3D-SAPCA with otherfilters
Department of Signal Processing
79
Comparison of BM3D-SAPCA with otherfilters (PSNR, SSIM)
Original Noisy, σ = 35 BM3D (27.82, 0.8207)
P.SADCT (27.51, 0.8143) SA-BM3D (28.02, 0.8228) BM3D-SAPCA (28.16, 0.8269)
Department of Signal Processing
80
Interpolation for Bayer Pattern
Original scene
Color Filter Array
Observation
Color Interpolation
Department of Signal Processing
81
Competitiveness with state-of-the-arttechniques
The proposed CFAI technique adapts to spatial properties of an image
Department of Signal Processing
82
Conventional Approach for Noiseless Data(Hamilton-Adams)
Department of Signal Processing
83
Karen Egiazarian
Proposed Approach for Noiseless Data(Spatially-Adaptive LPA-ICI)
Department of Signal Processing
84
Karen Egiazarian
Simulation of Radon reconstruction from sparse projections(approximating Radon projections as radial lines in FFT domain:Sparse projections: 11 radial lines)
Compressed Sensing Image Reconstruction viaRecursive BM3DEgiazarian, K., A. Foi, and V. Katkovnik, “Compressed Sensing Image Reconstruction viaRecursive Spatially Adaptive Filtering, ICIP 2007
Department of Signal Processing
Compressed Sensing Image Reconstructionvia Recursive BM3DEgiazarian, K., A. Foi, and V. Katkovnik, “Compressed Sensing ImageReconstruction via Recursive Spatially Adaptive Filtering, ICIP 2007
85
Karen Egiazarian
Simulation of Radon reconstruction from sparse projections(approximating Radon projections as Limited-angle in FFTdomain)
Department of Signal Processing
86
BM3D for upsampling and super-resolution
Image upsampling or zooming, can be de.ned as the process of resampling asingle low-resolution (LR) image on a high-resolution grid.
The process of combining a sequence of undersampled and degraded low-resolution images in order to produce a single high-resolution image is commonlyreferred to as a Super-resolution (SR) reconstruction.
Modern SR methods (e.g., Protter et al. 2008, Ebrahimi and Vrscay 2008) arebased on the nonlocal means (NLM) filtering paradigm (Buades-Coll-Morel,2005).• No explicit registration: one-to-one pixel mapping between frames is replaced bya one-to-many mapping.The BM3D and V-BM3D algorithms share with the NLM the idea of exploitingnonlocal similarity between blocks. However, in (V-)BM3D a more powerfultransform-domain modeling is used.
Department of Signal Processing
87
BM3D based superresolution
Department of Signal Processing
88
Image upsampling x 4(pixel replication)
Department of Signal Processing
89
Image upsampling x 4 in wavelet domain(Danielyan et al. EUSIPCO 2008)
Department of Signal Processing
90
Video superresolutioncomparison with (Protter et. al.)
1. M. Protter, M. Elad, H. Takeda, and P. Milanfar, .Generalizing the Non-Local-Means toSuper-Resolution Reconstruction., IEEE Trans. Image Process., 2008.
2. A. Danielyan, A. Foi, V. Katkovnik, and K. Egiazarian, .Image upsampling via spatiallyadaptive block-matching filtering, EUSIPCO2008, Lausanne, Switzerland, Aug. 2008.
Department of Signal Processing
91
Examples: Video denoising using V-BM3D
Department of Signal Processing
92
Examples: Video denoising using V-BM3D
Department of Signal Processing
93
Examples: Video denoising using V-BM3D
Department of Signal Processing
94
Conclusions
Our algorithms have been licensed to major digital cameramanufacturers and are already in use by various researchinstitutes for processing and enhancing their images.
Tomographic reconstruction of mouse embryo with BM3D filtering of axial slices(Harvard Medical School, Boston MA, 2010)
BM3D
Department of Signal Processing
95
Conclusions
Department of Signal Processing
96
Thanks to collaborators and former students:
Jaakko Astola (TUT, Finland)Leonid Yaroslavsky (Israel) –Sliding
DCT denoising (1997-2000)Dmitiy Paliy (NOKIA, Finland) LPA-ICI CFAI(2005-2007)Rusen Oktem (Berkeley, USA) Sliding DCTAram Danielyan (Noiseless Imaging)Enrique Sánchez-Monge (Noiseless Imaging)super-resolution using BM3D (2008-)Alessandro Foi (TUT, Finland)SA DCT, BM3D,…Vladimir Katkovnik (TUT, Finland) LPA-ICI,SA-DCT, BM3D,…Kostadin Dabov (Apple, USA),BM3DDmytro Rusanovsky (LG, USA)Video BM3Dand many-many others…