Mutually Guided Image Filtering

Image filtering is helpful to numerous multimedia, computer vision and graphics tasks. Linear translation-invariant filters with manually designed kernels have been widely used. However, their performance suffers from the content-blindness, say identically treating noises, textures and structures. To mitigate the content-blindness, a family of filters, called joint/guided filters, has attracted much attention from the community, the principle of which is transferring the structure in the reference image to the target one. The main drawback of most joint/guided filters comes from the ignorance of structural inconsistency between the reference and target signals that can be like color, infrared, and depth images captured under different conditions. Simply adopting such guidances very likely leads to unsatisfactory results. To address the above issues, we design a simple yet effective filter, named mutually guided image filter (muGIF), which jointly preserves mutual structures, avoids misleading from inconsistent structures, and smooths flat regions. Our muGIF is very flexible, which can perform in one of dynamic only (self-guided), static/dynamic and dynamic/dynamic modes.

Texture removal: RGF (b) and our muGIF (c) remove rich textures from (a). No-flash/flash image restoration: Guided by the flash image, the restored no-flash results of GIF and muGIF are (e) and (f), respectively. Mutual structure extraction: From the noisy depth/RGB (g), the mutual structures extracted by JFMS and muGIF are given in (h) and (i), respectively.

The demo program written in Matlab can be accessed from the following link in the form of a .rar file. Demo code

This demo software is provided for research purposes only. A license must be obtained for any commercial applications.

Related Works

Xiaojie Guo, Yu Li, Jiayi Ma, and Haibin Ling, "Mutually Guided Image Filtering" TPAMI, in press [PDF]

Xiaojie Guo, Yu Li, and Jiayi Ma, "Mutually Guided Image Filtering" ACM MM 17 [PDF]

Xiaojie Guo et al., "Structure-Texture Decomposition Via Joint Structure Discovery and Texture Smoothing" ICME 2018 [Project]]


We would like to thank the colleagues who collect the datasets used in this work. Considering the copyright issue, instead of collecting these data as a new dataset, we give the links to the datasets as follows:

Texture Smoothing:

Scale Space Filtering:

RGB/Depth Restoration:

RGB/NIR & Flash/No-flash Restoration:

Mutual Structure Discovery:

[Return to X. Guo's Homepage]