Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward...
Transcript of Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward...
![Page 1: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/1.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution PyramidsZeev Farbman, Raanan Fattal and Dani LischinskiSIGGRAPH Asia Conference (2011)
presented by:
Julian Steil
supervisor:
Prof. Dr. Joachim Weickert
Fig. 1.1: Gradient integration example
Fig. 1.2: Reconstruction result of Fig. 1.1
Seminar - Milestones and Advances in Image AnalysisProf. Dr. Joachim Weickert, Oliver DemetzMathematical Image Analysis GroupSaarland University
13th of November, 2012
![Page 2: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/2.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Overview
1. Motivation
2. Convolution Pyramids
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient Integration
6. Summary
![Page 3: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/3.jpg)
Motivation
Convolution
Gaussian Pyramid
Gaussian Pyramid -
Example
From Gaussian to Laplacian
Pyramid
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Overview
1. MotivationConvolutionGaussian PyramidGaussian Pyramid - ExampleFrom Gaussian to Laplacian Pyramid
2. Convolution Pyramids
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient Integration
6. Summary
2 / 22
![Page 4: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/4.jpg)
Motivation
Convolution
Gaussian Pyramid
Gaussian Pyramid -
Example
From Gaussian to Laplacian
Pyramid
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Motivation
Convolution
Two-Dimensional Convolution:• discrete convolution of two imagesg = (gi,j)i,j∈Z and w = (wi,j)i,j∈Z :
(g ∗ w)i,j :=∑k∈Z
∑`∈Z
gi−k,j−`wk,` (1)
• components of convolution kernel w can be regarded as mirroredweights for averaging the components of g
• the larger the kernel size the larger the runtime• ordinary convolution implementation needs O(n2)
3 / 22
![Page 5: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/5.jpg)
Motivation
Convolution
Gaussian Pyramid
Gaussian Pyramid -
Example
From Gaussian to Laplacian
Pyramid
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Motivation
Gaussian Pyramid
• sequence of images g0, g1, ..., gn• computed by a filtering procedure equivalent to convolution with a
local, symmetric weighting function=⇒ e.g. a Gaussian kernel
Procedure:• image initialised by array g0 which contains C columns and R rows• each pixel represents the light intensity I between 0 and 255
=⇒ g0 is the zero level of Gaussian Pyramid• each pixel value in level i is computed as a weighting average of
level i− 1 pixel values
Fig. 2: One-dimensional graphic representation of the Gaussian pyramid4 / 22
![Page 6: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/6.jpg)
Motivation
Convolution
Gaussian Pyramid
Gaussian Pyramid -
Example
From Gaussian to Laplacian
Pyramid
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Motivation
Gaussian Pyramid - Example
Fig. 3: First six levels of the Gaussian pyramid for the “Lena” image. The original image, level 0, measures 257x257 pixels =⇒ level 5 measures just 9x9 pixels
Remark:
density of pixels is reduced by half in one dimension and by fourth intwo dimensions from level to level
5 / 22
![Page 7: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/7.jpg)
Motivation
Convolution
Gaussian Pyramid
Gaussian Pyramid -
Example
From Gaussian to Laplacian
Pyramid
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Motivation
From Gaussian to Laplacian Pyramid
Fig. 4: First four levels of the Gaussian and Laplacian pyramid of Fig.3.
• each level of Laplacian pyramid is the difference between thecorresponding and the next higher level of the Gaussian pyramid
• full expansion is used in Fig. 4 to help visualise the contents thepyramid images
6 / 22
![Page 8: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/8.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Overview
1. Motivation
2. Convolution PyramidsApproachForward and Backward TransformFlow Chart and PseudocodeOptimisation
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient Integration
6. Summary
6 / 22
![Page 9: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/9.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Approach
Task:• approximate effect of convolution with large kernels
=⇒ higher spectral accuracy + translation-invariant operation• Is it also possible in O(n)?
Idea:• use of repeated convolution with small kernels on multiple scales• disadvantage: not translation-invariant due to subsampling
operation to reach O(n) performance
Method:• pyramids rely on a spectral “divide-and-conquer” strategy• no subsampling of the decomposed signal increases the
translation-invariance• use finite impulse response filters to achieve some spacial
localisation and runtime O(n)
7 / 22
![Page 10: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/10.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Approach
Task:• approximate effect of convolution with large kernels
=⇒ higher spectral accuracy + translation-invariant operation• Is it also possible in O(n)?
Idea:• use of repeated convolution with small kernels on multiple scales• disadvantage: not translation-invariant due to subsampling
operation to reach O(n) performance
Method:• pyramids rely on a spectral “divide-and-conquer” strategy• no subsampling of the decomposed signal increases the
translation-invariance• use finite impulse response filters to achieve some spacial
localisation and runtime O(n)
7 / 22
![Page 11: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/11.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Approach
Task:• approximate effect of convolution with large kernels
=⇒ higher spectral accuracy + translation-invariant operation• Is it also possible in O(n)?
Idea:• use of repeated convolution with small kernels on multiple scales• disadvantage: not translation-invariant due to subsampling
operation to reach O(n) performance
Method:• pyramids rely on a spectral “divide-and-conquer” strategy• no subsampling of the decomposed signal increases the
translation-invariance• use finite impulse response filters to achieve some spacial
localisation and runtime O(n)
7 / 22
![Page 12: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/12.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Forward and Backward Transform
Forward Transform - Analysis Step:
• convolve a signal with a first filter h1
• subsample the result by a factor of two• process is repeated on the subsampled data• an unfiltered and unsampled copy of the signal is kept at each level
al0 = al (2)
al+1 = ↓ (h1 ∗ al) (3)
Backward Transform - Synthesis Step:
• upsample by inserting a zero between every two samples• convolve the result with a second filter h2
• combine upsampled signal with the signal stored at each level afterconvolving with a third filter g
al = h2 ∗ (↑ al+1) + g ∗ al0 (4)
8 / 22
![Page 13: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/13.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Forward and Backward Transform
Forward Transform - Analysis Step:
• convolve a signal with a first filter h1
• subsample the result by a factor of two• process is repeated on the subsampled data• an unfiltered and unsampled copy of the signal is kept at each level
al0 = al (2)
al+1 = ↓ (h1 ∗ al) (3)
Backward Transform - Synthesis Step:
• upsample by inserting a zero between every two samples• convolve the result with a second filter h2
• combine upsampled signal with the signal stored at each level afterconvolving with a third filter g
al = h2 ∗ (↑ al+1) + g ∗ al0 (4)
8 / 22
![Page 14: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/14.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Flow Chart and Pseudocode
Fig. 5: Flow Chart to visualise pyramid structure, source taken from [1]
Algorithm 1 Multiscale Transform1: Determine the number of levels L
2: Forward transform (analysis)3: a0 = a
4: for each level l = 0...L− 1 do5: al0 = al
6: al+1 = ↓ (h1 ∗ al)7: end for8: Backward transform (synthesis)9: aL = g ∗ aL
10: for each level l = L− 1...0 do11: al = h2 ∗ (↑ al+1) + g ∗ al012: end for
9 / 22
![Page 15: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/15.jpg)
Motivation
Convolution Pyramids
Approach
Forward and Backward
Transform
Flow Chart and
Pseudocode
Optimisation
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Convolution Pyramids
Optimisation
Kernel Determination:• target kernel f is given• seek a set of kernels F = h1, h2, g that minimise
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f︸︷︷︸targetkernel
∗ a︸︷︷︸inputsignal
‖ (5)
• kernels in F should be small and separable• use larger and/or non-separable filters increase accuracy
=⇒ specific choice depends on application requirements• remarkable results using separable kernels in F for non-separable
target filters f• target filters f with rotational and mirroring symmetries enforce
symmetry on h1, h2, g
10 / 22
![Page 16: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/16.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Overview
1. Motivation
2. Convolution Pyramids
3. Application 1 - Gaussian KernelsGaussian Kernel ConvolutionExample - Gaussian FilterExample - Scattered Data Interpolation
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient Integration
6. Summary
10 / 22
![Page 17: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/17.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Application 1 - Gaussian Kernels
Gaussian Kernel Convolution
Task:
• approximate Gaussian kernels e‖x‖2
2σ2 at the original fine grid inO(n)
• no truncated filter support
Determination of F = h1, h2, g:
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f︸︷︷︸target
Gaussiankernel
∗ a︸︷︷︸image
toconvolve
‖ (5)
Problem:
• Gaussians are rather efficient low-pass filters• pyramid contains high-frequent components coming from finer
levels introduced by convolution with g
11 / 22
![Page 18: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/18.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Application 1 - Gaussian Kernels
Gaussian Kernel Convolution
Task:
• approximate Gaussian kernels e‖x‖2
2σ2 at the original fine grid inO(n)
• no truncated filter support
Determination of F = h1, h2, g:
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f︸︷︷︸target
Gaussiankernel
∗ a︸︷︷︸image
toconvolve
‖ (5)
Problem:
• Gaussians are rather efficient low-pass filters• pyramid contains high-frequent components coming from finer
levels introduced by convolution with g
11 / 22
![Page 19: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/19.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Application 1 - Gaussian Kernels
Gaussian Kernel Convolution
Task:
• approximate Gaussian kernels e‖x‖2
2σ2 at the original fine grid inO(n)
• no truncated filter support
Determination of F = h1, h2, g:
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f︸︷︷︸target
Gaussiankernel
∗ a︸︷︷︸image
toconvolve
‖ (5)
Problem:
• Gaussians are rather efficient low-pass filters• pyramid contains high-frequent components coming from finer
levels introduced by convolution with g
11 / 22
![Page 20: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/20.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Application 1 - Gaussian Kernels
Example - Gaussian Filter
Solution:• modulation of g at each level l• higher wl at the levels closest to the target size• for different σ different sets of kernels F are necessary
Fig. 6.1: Original image,source: taken from [1]
Fig. 6.2: Exact convolution with a Gaussian filter(σ = 4), source: taken from [1]
Fig. 6.3: Convolution using optimizationapproach forσ = 4, source: taken from [1]
Fig. 7.1: Exact kernels (in red) withapproximated kernels (in blue),source: taken from [1]
Fig. 7.2: Exact Gaussian (red), approximationusing 5x5 kernels (blue) and 7x7 kernel(green) , source: taken from [1]
Fig. 7.3: Magnification of Fig. 7.2 shows betteraccuracy of larger kernels,source: taken from [1]
used kernels
12 / 22
![Page 21: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/21.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Gaussian Kernel
Convolution
Example - Gaussian Filter
Example - Scattered Data
Interpolation
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Application 1 - Gaussian Kernels
Example - Scattered Data Interpolation
Fig. 8.4: Approximation withwider Gaussian,source: taken from [1]
Fig. 8.5: Approximation withnarrower Gaussian,source: taken from [1]
Fig. 8.6: Exact resultscorresponding to red widerGaussian , source: taken from[1]
Fig. 8.7: Exact resultscorresponding to red narrowerGaussian,source: taken from [1]
Fig. 8.1: Horizontal slice through exactwider Gaussian (red) andapproximation (blue),source: taken from [1]
Fig. 8.2: Horizontal slice through exactnarrower Gaussian (red) andapproximation (blue),source: taken from [1]
Fig. 8.3: Scattered datainterpolation input ,source: taken from [1]
13 / 22
![Page 22: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/22.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
How to use boundary
interpolation?
Example - Seamless
Cloning
Application 3 -Gradient Integration
Summary
Overview
1. Motivation
2. Convolution Pyramids
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary InterpolationHow to use boundary interpolation?Example - Seamless Cloning
5. Application 3 - Gradient Integration
6. Summary
13 / 22
![Page 23: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/23.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
How to use boundary
interpolation?
Example - Seamless
Cloning
Application 3 -Gradient Integration
Summary
Application 2 - Boundary Interpolation
How to use boundary interpolation?
Seamless Image Cloning:
• formulation as boundary value problem• effectively solved by constructing a smooth membrane• interpolation of differences along a seam between two images
Shepard’s Method:
• Ω is region of interest and boundary values are given by b(x)
• smoothly interpolates boundary values to all grid points inside Ω
• defines interpolant r at x as weighted average of boundary values:
r(x) =
∑k wk(x)b(xk)∑
k wk(x)=⇒ r(xi) =
∑nj=0 w(xi, xj)r(xj)∑nj=0 w(xi, xj)χr(xj)
=w ∗ rw ∗ χr
(6)
• xk = boundary points, b(xk) = boundary values• weight function wk(x) is given by
wk(x) = w(xk, x) =1
d(xk, x)3(7)
• strong spike at xk and decays rapidly away from it• computational cost O(Kn), K boundary values and n points in Ω
14 / 22
![Page 24: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/24.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
How to use boundary
interpolation?
Example - Seamless
Cloning
Application 3 -Gradient Integration
Summary
Application 2 - Boundary Interpolation
How to use boundary interpolation?
Seamless Image Cloning:
• formulation as boundary value problem• effectively solved by constructing a smooth membrane• interpolation of differences along a seam between two images
Shepard’s Method:
• Ω is region of interest and boundary values are given by b(x)
• smoothly interpolates boundary values to all grid points inside Ω
• defines interpolant r at x as weighted average of boundary values:
r(x) =
∑k wk(x)b(xk)∑
k wk(x)=⇒ r(xi) =
∑nj=0 w(xi, xj)r(xj)∑nj=0 w(xi, xj)χr(xj)
=w ∗ rw ∗ χr
(6)
• xk = boundary points, b(xk) = boundary values• weight function wk(x) is given by
wk(x) = w(xk, x) =1
d(xk, x)3(7)
• strong spike at xk and decays rapidly away from it• computational cost O(Kn), K boundary values and n points in Ω
14 / 22
![Page 25: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/25.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
How to use boundary
interpolation?
Example - Seamless
Cloning
Application 3 -Gradient Integration
Summary
Application 2 - Boundary Interpolation
Example - Seamless Cloning
Determination of F = h1, h2, g:
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f ∗ a︸ ︷︷ ︸exact
membraner(x)
‖ (5)
Fig. 9.1: Source image,source: taken from [2]
Fig. 9.2: Membrane mask,source: taken from [2]
Fig. 9.3: Target image,source: taken from [2]
Fig. 9.4: Approximated membranesource: taken from [1]
Fig. 9.5: Superimposed image with a clonedpatch, source: taken from [1]
Fig. 9.6: Result of applying Fig. 9.4 to Fig. 9.5,source: taken from [1]
Used Kernels
15 / 22
![Page 26: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/26.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Overview
1. Motivation
2. Convolution Pyramids
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient IntegrationKernel DetectionExample - Gradient IntegrationHow does the target filter look like?Reconstruction of Target Filter
6. Summary
15 / 22
![Page 27: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/27.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Application 3 - Gradient Integration
Kernel Detection
Determination of F = h1, h2, g:• choose a natural image I• a is the divergence of its gradient field:
a = div∇I (8)
I = f ∗ a (9)
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f ∗ a︸ ︷︷ ︸naturalimage
I
‖ (5)
Fig. 10.1: Natural image I ,source: taken from [1]
Fig. 10.2: Corresponding gradient imagea ofFig. 10.1, source: taken from [1]
16 / 22
![Page 28: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/28.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Application 3 - Gradient Integration
Example - Gradient Integration
Fig. 11.1: Gradient image of Fig 11.4,source: taken from [1]
Fig. 11.2: Reconstruction of Fig. 11.1 withF5,3 , source: taken from [1]
Fig. 11.3: Reconstruction of Fig. 11.1 withF7,5 , source: taken from [1]
Fig. 11.4: Original image (512x512),source: taken from [1]
Fig. 11.5: Absolute errors of Fig. 11.2(magnified by x50), source: taken from [1]
Fig. 11.6: Absolute errors of Fig. 11.3(magnified by x50), source: taken from [1]
Used Kernels
17 / 22
![Page 29: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/29.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Application 3 - Gradient Integration
How does the target filter look like?
Task:
• recover image u (here: u = a0F ) by solving the Poisson equation
4u = div v (10)• v = gradient field
Solution:
• Green’s functions
G(x, x′) = G(‖x− x′‖) =1
2πlog
1
‖x− x′‖ (11)
define fundamental solutions to the Poisson equation
4G(x, x′) = δ(x, x′) (12)
• δ = discrete delta function• (10) is defined over an infinite domain with no boundary constraints
=⇒ Laplace operator becomes spatially invariant=⇒ Green’s function becomes translation invariant
• solution of (10) is given by the convolution
u = G ∗ div v (13)18 / 22
![Page 30: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/30.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Application 3 - Gradient Integration
How does the target filter look like?
Task:
• recover image u (here: u = a0F ) by solving the Poisson equation
4u = div v (10)• v = gradient field
Solution:
• Green’s functions
G(x, x′) = G(‖x− x′‖) =1
2πlog
1
‖x− x′‖ (11)
define fundamental solutions to the Poisson equation
4G(x, x′) = δ(x, x′) (12)
• δ = discrete delta function• (10) is defined over an infinite domain with no boundary constraints
=⇒ Laplace operator becomes spatially invariant=⇒ Green’s function becomes translation invariant
• solution of (10) is given by the convolution
u = G ∗ div v (13)18 / 22
![Page 31: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/31.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Kernel Detection
Example - Gradient
Integration
How does the target filter
look like?
Reconstruction of Target
Filter
Summary
Application 3 - Gradient Integration
Reconstruction of Target Filter
Target Filter Determination:
• using results of previous F = h1, h2, g• a is a centered delta function
a = div∇I (8)
I = f ∗ a (9)
• Green’s function provides a suitable result for f
Fig. 12.1: Reconstruction of the Green’s function,source: taken from [1]
Fig. 12.2: space invariant corresponding kernel of Fig. 12.1,source: taken from [1]
19 / 22
![Page 32: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/32.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Summary
Overview
1. Motivation
2. Convolution Pyramids
3. Application 1 - Gaussian Kernels
4. Application 2 - Boundary Interpolation
5. Application 3 - Gradient Integration
6. SummarySummary
19 / 22
![Page 33: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/33.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
Summary
Summary
Summary
• approximation of large convolution filters in O(n)
=⇒ using kernels of small support F = h1, h2, g+ multiscale pyramid scheme
• kernel determination by optimization:
arg minF
‖ a0F︸︷︷︸result of
multiscaletransform
− f︸︷︷︸targetkernel
∗ a︸︷︷︸inputsignal
‖
• suitable for different applications like...• gradient integration• seamless cloning• scattered data interpolation
20 / 22
![Page 34: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/34.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
References
References
[1] ZEEV FARBMAN, RAANAN FATTAL, DANI LISCHINSKI
Convolution pyramidsProc. 2011 SIGGRAPH Asia Conference, Article No. 175The Hebrew University (2011)
[2] COMPUTER GRAPHICS & COMPUTATIONAL PHOTOGRAPHY LAB
Supplementary Materials of the paper “Convolution pyramids”The Hebrew University (2011)http://www.cs.huji.ac.il/labs/cglab/projects/convpyr/
[3] MATHEMATICAL IMAGE ANALYSIS GROUP
Lecture notes of the “Image Processing and Computer Vision” lectureSaarland University. Winter term (2011)http://www.mia.uni-saarland.de/Teaching/ipcv06.shtml
[4] PETER J. BURT, EDWARD H. ADELSON
The Laplacian Pyramid as a Compact Image CodeIEEE Transcriptions on CommunicationsVol. COM-31, No. 4, (April 1983)
21 / 22
![Page 35: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/35.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
References
Thank you for your attention!
22 / 22
![Page 36: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/36.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
I
![Page 37: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/37.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
II
![Page 38: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/38.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
III
![Page 39: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/39.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
IV
![Page 40: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/40.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
V
![Page 41: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/41.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
VI
![Page 42: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/42.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
VII
![Page 43: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/43.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
VIII
![Page 44: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/44.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
IX
![Page 45: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/45.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
X
![Page 46: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/46.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XI
![Page 47: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/47.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XII
![Page 48: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/48.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XIII
![Page 49: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/49.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XIV
![Page 50: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/50.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XV
![Page 51: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/51.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XVI
![Page 52: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/52.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XVII
![Page 53: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/53.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XVIII
![Page 54: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/54.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XIX
![Page 55: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/55.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XX
![Page 56: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/56.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXI
![Page 57: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/57.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXII
![Page 58: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/58.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXIII
![Page 59: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/59.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXIV
![Page 60: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/60.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXV
![Page 61: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/61.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXVI
![Page 62: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/62.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXVII
![Page 63: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/63.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXVIII
![Page 64: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/64.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXIX
![Page 65: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/65.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXX
![Page 66: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/66.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXXI
![Page 67: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/67.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXXII
![Page 68: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/68.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXXIII
![Page 69: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/69.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXXIV
![Page 70: Convolution Pyramids - uni-saarland.de · Convolution Pyramids Approach Forward and Backward Transform Flow Chart and Pseudocode Optimisation Application 1 - Gaussian Kernels Application](https://reader030.fdocuments.in/reader030/viewer/2022040916/5e8fa26b5693b67b09277c6b/html5/thumbnails/70.jpg)
Motivation
Convolution Pyramids
Application 1 -Gaussian Kernels
Application 2 -Boundary Interpolation
Application 3 -Gradient Integration
Summary
XXXV