Back to Jue Wang's project page

Noise Brush:

Interactive High Quality Image-Noise Separation


Given an input noisy image (a) and an initial denoising result (b) consisting of a latent image layer (top) and a noise layer (bottom) (this example is produced by Noiseware [Imagenomic Inc. 2008]), our system provides the user with a set of easy interactive control to achieve high quality image-noise separation shown in (c). The close-up comparisons show that our method can effectively restore from (d) high-frequency image structures, as shown in (e) where a sharper latent image is obtained.


Jia Chen*,  Chi-Keung Tang            The Hong Kong University of Science and Technology
Jue Wang                                           Adobe Systems

*Jia did most of the work when he was an intern at Adobe.


This paper proposes an interactive approach using joint imagenoise filtering for achieving high quality image-noise separation. The core of the system is our novel joint image-noise filter which operates in both image and noise domain, and can effectively separate noise from both high and low frequency image structures. A novel user interface is introduced, which allows the user to interact with both the image and the noise layer, and apply the filter adaptively and locally to achieve optimal results. A comprehensive and quantitative evaluation shows that our interactive system can significantly improve the initial image-noise separation results. Our system can also be deployed in various noise-consistent image editing tasks, where preserving the noise characteristics inherent in the input image is a desired feature.


                                                                                     PDF (26 MB)

Additional Slides

                                                         71-page slides in PDF format (31.4 MB)


                                                                      QuickTime MP4 (72.4 MB)


                                                                                       Stay tuned.