Download presentation

Presentation is loading. Please wait.

Published byBathsheba Boyd Modified over 2 years ago

2
Accelerating Spatially Varying Gaussian Filters Jongmin Baek and David E. Jacobs Stanford University

3
Motivation Input Gaussian Filter Spatially Varying Gaussian Filter

4
1) Accelerating Spatially Varying Gaussian Filters 2) Accelerating Spatially Varying Gaussian Filters 3) Accelerating Spatially Varying Gaussian Filters 4) Applications Roadmap

5
Gaussian Filters Position Value

6
Gaussian Filters Each output value …

7
Gaussian Filters … is a weighted sum of input values …

8
Gaussian Filters … whose weight is a Gaussian …

9
Gaussian Filters … in the space of the associated positions.

10
Gaussian Blur Gaussian Filters: Uses

11
Bilateral Filter Gaussian Filters: Uses

12
Non-local Means Filter Gaussian Filters: Uses

13
Applications Denoising images and meshes Data fusion and upsampling Abstraction / Stylization Tone-mapping ... Gaussian Filters: Summary Previous work on fast Gaussian Filters Bilateral Grid (Chen, Paris, Durand; 2007) Gaussian KD-Tree (Adams et al.; 2009) Permutohedral Lattice (Adams, Baek, Davis; 2010)

14
Summary of Previous Implementations: A separable blur flanked by resampling operations. Exploit the separability of the Gaussian kernel. Gaussian Filters: Implementations

15
Spatially Varying Gaussian Filters Spatially varying covariance matrix Spatially Invariant

16
Trilateral Filter (Choudhury and Tumblin, 2003) Tilt the kernel of a bilateral filter along the image gradient. “Piecewise linear” instead of “Piecewise constant” model. Spatial Variance in Previous Work

17
Spatially Varying Gaussian Filters: Tradeoff Benefits: Can adapt the kernel spatially. Better filtering performance. Cost: No longer separable. No existing acceleration schemes. Input Bilateral-filtered Trilateral-filtered

18
Problem: Spatially varying (thus non-separable) Gaussian filter Existing Tool: Fast algorithms for spatially invariant Gaussian filters Solution: Re-formulate the problem to fit the tool. Need to obey the “piecewise-constant” assumption Acceleration

19
Naïve Approach (Toy Example) I LOST THE GAME Input Signal Desired Kernel 1 11234 filtered w/ 1 filtered w/ 2 filtered w/ 3 filtered w/ 4 111 2 3 Output Signal 4

20
In practice, the # of kernels can be very large. Challenge #1 Pixel Location x Desired Kernel K(x) Range of Kernels needed

21
Sample a few kernels and interpolate. Solution #1 Desired Kernel K(x) Sampled kernels Interpolate result! Pixel Location x K1K1 K2K2 K3K3

22
Interpolation needs an extra assumption to work: The covariance matrix Ʃ i is either piecewise- constant, or smoothly varying. Kernel is spatially varying, but locally spatially invariant. Assumptions

23
Runtime scales with the # of sampled kernels. Challenge #2 Desired Kernel K(x) Filter only some regions of the image with each kernel. (“support”) Pixel Location x Sampled kernels K1K1 K2K2 K3K3

24
In this example, x needs to be in the support of K 1 & K 2. Defining the Support Desired Kernel K(x) Pixel Location x K1K1 K2K2 K3K3

25
Dilating the Support Desired Kernel K(x) Pixel Location x K1K1 K2K2 K3K3

26
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results.

27
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results. K1K1 K2K2 K3K3

28
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results. K1K1 K2K2 K3K3

29
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results. K1K1 K2K2 K3K3

30
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results. K1K1 K2K2 K3K3

31
Algorithm 1) Identify kernels to sample. 2) For each kernel, compute the support needed. 3) Dilate each support. 4) Filter each dilated support with its kernel. 5) Interpolate from the filtered results. K1K1 K2K2 K3K3

32
Applications HDR Tone-mapping Joint Range Data Upsampling

33
Application #1: HDR Tone-mapping Input HDR Detail Base Filter Output Attenuate

34
Tone-mapping Example Bilateral Filter Kernel Sampling

35
Application #2: Joint Range Data Upsampling Range Finder Data Sparse Unstructured Noisy Scene Image Output Filter

36
Synthetic Example Scene Image Ground Truth Depth

37
Synthetic Example Scene ImageSimulated Sensor Data

38
Synthetic Example : Result Kernel Sampling Bilateral Filter

39
Synthetic Example : Relative Error Bilateral Filter Kernel Sampling 2.41% Mean Relative Error0.95% Mean Relative Error

40
Real-World Example Scene Image Range Finder Data *Dataset courtesy of Jennifer Dolson, Stanford University

41
Real-World Example: Result Input Bilateral Naive Kernel Sampling

42
Performance Kernel Sampling Choudhury and Tumblin (2003) Naïve Tonemap1 5.10 s41.54 s312.70 s Tonemap2 6.30 s88.08 s528.99 s Kernel Sampling (No segmentation) Depth1 3.71 s57.90 s Depth2 9.18 s131.68 s

43
1.A generalization of Gaussian filters Spatially varying kernels Lose the piecewise-constant assumption. 2.Acceleration via Kernel Sampling Filter only necessary pixels (and their support) and interpolate. 3.Applications Conclusion

Similar presentations

OK

A Gentle Introduction to Bilateral Filtering and its Applications 07/10: Novel Variants of the Bilateral Filter Jack Tumblin – EECS, Northwestern University.

A Gentle Introduction to Bilateral Filtering and its Applications 07/10: Novel Variants of the Bilateral Filter Jack Tumblin – EECS, Northwestern University.

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google

Ppt on fuel from plastic waste Class-9th physics ppt on motion Ppt on game theory mario Ppt on conference call etiquette Ppt on australian continent countries Ppt on obstructive sleep apnea Ppt on grid connected pv system Ppt on scientific calculator project in java Ppt on file security system Ppt on preservation of public property records