Example of applications of domain transform edge-aware filters.
Domain Transform for Edge-Aware Image and Video Processing
Eduardo S. L. Gastal
eslgastal@inf.ufrgs.br
and Manuel M. Oliveira
oliveira@inf.ufrgs.br

Instituto de Informática, UFRGS

Logo INF Logo UFRGS

ACM Transactions on Graphics.
Volume 30 (2011), Number 4, Proceedings of SIGGRAPH 2011, Article 69.


Contents

Abstract Downloads Results FAQ Reference Acknowledgements

Abstract

We present a new approach for performing high-quality edge-preserving filtering of images and videos in real time. Our solution is based on a transform that defines an isometry between curves on the 2D image manifold in 5D and the real line. This transform preserves the geodesic distance between points on these curves, adaptively warping the input signal so that 1D edge-preserving filtering can be efficiently performed in linear time. We demonstrate three realizations of 1D edge-preserving filters, show how to produce high-quality 2D edge-preserving filters by iterating 1D-filtering operations, and empirically analyze the convergence of this process. Our approach has several desirable features: the use of 1D operations leads to considerable speedups over existing techniques and potential memory savings; its computational cost is not affected by the choice of the filter parameters; and it is the first edge-preserving filter to work on color images at arbitrary scales in real time, without resorting to subsampling or quantization. We demonstrate the versatility of our domain transform and edge-preserving filters on several real-time image and video processing tasks including edge-preserving filtering, depth-of-field effects, stylization, recoloring, colorization, detail enhancement, and tone mapping.

Downloads

NEW: Learn how to obtain practically unlimited control over the shape of the filtering kernel when using the Domain Transform! Refer to our work presented at Eurographics 2015:

"High-Order Recursive Filtering of Non-Uniformly Sampled Signals for Image and Video Processing".

http://inf.ufrgs.br/~eslgastal/NonUniformFiltering/

Paper

Full paper (3.4 MB, 11 pages)
JPEG compressed images.

ACM Copyright Notice: © ACM, 2011. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version was published in ACM TOG, Vol. 30, 4, July 2011 — http://doi.acm.org/10.1145/1964921.1964964.

Source Code

Domain Transform edge-preserving filters source code (0.7 MB)
Version 1.0 for MATLAB/GNU Octave.

Demo

Domain Transform edge-preserving filters demo (5.4 MB)
Version 1.0 for Windows 32-bit/64-bit and Linux 64-bit.

Screenshot:

Results

Video

Watch on YouTube  /  Download WMV (17 MB)

Lamp Images

Lamp Photograph Lamp Edge-Aware Smoothing Lamp Detail Enhancement Lamp Stylization
Input Photograph Edge-aware smoothing Detail enhancement Stylization

Lamp Recoloring Lamp Pencil Drawing Lamp Depth-of-field
Recoloring Pencil-drawing Depth-of-field

1080p Real-Time Video Filtering

Input Edge-aware smoothing Stylization Pencil-drawing
Big Buck Bunny Input Example 0 Big Buck Bunny Smoothing Example 0 Big Buck Bunny Stylization Example 0 Big Buck Bunny Pencil Example 0
Big Buck Bunny Input Example 1 Big Buck Bunny Smoothing Example 1 Big Buck Bunny Stylization Example 1 Big Buck Bunny Pencil Example 1
Big Buck Bunny Input Video
Download AVI
(11.7 MB)
Big Buck Bunny Smoothing Video
Download AVI
(7.8 MB)
Big Buck Bunny Stylization Video
Download AVI
(14 MB)
Big Buck Bunny Pencil Video
Download MP4
(9.8 MB)
Big Buck Bunny Input Video
Watch on YouTube
Big Buck Bunny Smoothing Video
Watch on YouTube
Big Buck Bunny Stylization Video
Watch on YouTube
Big Buck Bunny Pencil Video
Watch on YouTube

Big Buck Bunny input clip and images © copyright 2008, Blender Foundation / www.bigbuckbunny.org.


More Applications of our Real-Time Edge-Aware Filters

Important: The following materials are best viewed in a web browser that is not Internet Explorer (IE), since it cannot resize the images properly. Furthermore, in IE you might need to allow ActiveX content to display the images.

Additional Materials

FAQ

Are the domain transform filters temporally coherent?

Yes, as long as the image derivatives are temporally coherent. In the video filtering examples above each frame was filtered independently and the results are of high quality without temporal artifacts.

Can the domain transform filters be applied for image denoising?

We have not performed an extensive evaluation, however we have had some success using the domain transform filters for removing small to medium amounts of noise. The main issue with noisy images is finding robust estimates for their derivatives, which cannot be done using simple forward differences. One possible solution is pre-filtering the image with a low-pass filter to avoid large oscillations in the image gradient. See [1] for a similar approach applied to anisotropic diffusion filtering.

[1] F. Catte, P.L. Lions, J.M. Morel, and T. Coll. Image selective smoothing and edge detection by nonlinear diffusion. SIAM Journal of Numerical Analysis, 29(1):182-193, 1992.

Reference

Citation

Eduardo S. L. Gastal and Manuel M. Oliveira. "Domain Transform for Edge-Aware Image and Video Processing". ACM Transactions on Graphics. Volume 30 (2011), Number 4, Proceedings of SIGGRAPH 2011, Article 69.

BibTeX

@article{GastalOliveira2011DomainTransform,
  author    = {Eduardo S. L. Gastal and Manuel M. Oliveira},
  title     = {Domain Transform for Edge-Aware Image and Video Processing},
  journal   = {ACM TOG},
  volume    = {30},
  number    = {4},
  year      = {2011},
  pages     = {69:1--69:12},
  articleno = {69},
  note      = {Proceedings of SIGGRAPH 2011}
}
  

Keywords

Domain Transform, Edge-Preserving Filtering, Anisotropic Diffusion, Bilateral Filter.

Acknowledgements

CNPq-Brazil fellowships and grants # 150550/2015-4, 158666/2010-0, 557814/2010-3, 200284/2009-6, 308936/2010-8, and 480485/2010-0.

Last updated: July 13, 2015.