Cross-Field Joint Image Restoration via Scale Map

Qiong Yan^§      Xiaoyong Shen^§      Li Xu^§      Shaojie Zhuo^†
Xiaopeng Zhang^†    Liang Shen^†       Jiaya Jia^§

^§The Chinese Univeristy of Hong Kong       ^†Qualcomm Incorporated

An overview of our cross-field joint image restoration framework. Given image pairs in (a) and (b), our method can get the high-quality restoration result in (c). The s maps in different iterations are shown in (e)-(h).

Abstract

Color, infrared and flash images captured in different fields can be employed to effectively eliminate noise and other visual artifacts. We propose a two-image restoration framework considering input images from different fields, for example, one noisy color image and one dark-flashed near-infrared image. The major issue in such a framework is to handle all structure divergence and find commonly usable edges and smooth transitions for visually plausible image reconstruction. We introduce a novel scale map as a competent representation to explicitly model derivative-level confidence and propose new functions and a numerical solver to effectively infer it following our important structural observations. Our method is general and shows the principled way to solve cross-field restoration problems.

Results

Download ALL high quality images

RGB/flashed-NIR restoration
Noisy Input	NIR Image
Click buttons to show results made by different algorithms
Original   BM3D   Guided Image Filter   Our Result

RGB/flashed-NIR restoration
Non-flash Image [1]	Flash Image [1]
Click buttons to show results made by different algorithms
Original   Guided Image Filter   Petschnigg et al.[2004]   Our Result

Day/night image restoration
Day Image [2]	Night Image [2]
Click buttons to show different results
Original   PhotoShop   Our Result

Haze Removal
Haze Image [3]	NIR Image [3]
Click buttons to show different results
Original   Schaul et. al. [2009]   He et. al. [2011]   Our Result

RGB-Depth image restoration
RGB Image	Raw Depth Image with Noise
Click buttons to show different results
Original   Our Result

Downloads

"Cross-Field Joint Image Restoration via Scale Map" Qiong Yan, Xiaoyong Shen, Li Xu, Shaojie Zhuo, Xiaopeng Zhang, Liang Shen, Jiaya Jia IEEE International Conference on Computer Vision(ICCV), 2013   [Paper (pdf, 2.90MB)] [BibTex]   [Supplementary file (pdf, 13.5MB)]  [Matlab executable (zip, 2.5MB)]   [Results (zip, 35.2MB)]

References

[1] G. Petschnigg, R. Szeliski, M. Agrawala, M. F. Cohen, H. Hoppe, and K. Toyama. Digital photography with flash and no-flash image pairs. ACM Trans. Graph., 23(3):664-672,2004.

[2] R. Raskar, A. Ilie, and J. Yu. Image fusion for context enhancement and video surrealism. In NPAR, pages 85–152, 2004.

[3] L. Schaul, C. Fredembach, and S. Susstrunk. Color image dehazing using the near-infrared. In ICIP, pages 1629–1632, 2009.

Last update: Oct. 4, 2015