Single Image Defogging by Multi-level Depth Inference and Airlight Estimation


Authors: Dr. Yuan-Kai Wang, Ching-Tang Fan, and Chia-Wei Chang

Department of Electrical Engineering Fu Jen Catholic University

ABSTRACT

This paper presents an automatic method for the defogging from a single haze image. To recover a foggy image, an accurate depth map is estimated from a multi-level inference method, which fuses depth maps with different sizes of patches by dark channel prior. Markov random field (MRF) is applied to label the depth level in adjacent region for the compensation of wrong estimated regions. Airlight is automatically estimated as the deepest and largest area from the MRF labeled result. The accurate estimation of airlight provides good restoration with respect to visibility and contrast but without oversaturating. The algorithm is verified by a handful of foggy and hazy images. Experimental results demonstrate that the defogging method can recover high-quality images through accurate estimation of depth map and airlight.


Related publications:

    Y. K. Wang, C. T. Fan, and C. W. Chang, "Accurate Depth Estimation for Image Defogging using Markov Random Field," in 4th International Conference on Graphic and Image Processing, Singapore, 2012

    Y. K. Wang, C. T. Fan, and C. W. Chang, "Single Image Defogging by Multi-level Depth Estimation and Automatic Airlight Extraction," The 25th IPPR Conference on Computer Vision, Graphics, and Image Processing, Taiwan, 2012


Other materials:

    ICGIP 2012 presentation slides

    CVGIP 2012 presentation slides

    Experimental images


Results and Depth maps:

input foggy image

our result

our depth map

input foggy image

our result

our depth map

input foggy image

our result

our depth map

input foggy image

our result

our depth map


More Results:

input foggy image

our result

input foggy image

our result

input foggy image

our result

input foggy image

our result

input foggy image

our result


Iterative defogging:

input foggy image

first iteration result

second iteration result

third iteration result

fourth iteration result

fifth iteration result


Comparisons:

input foggy image

He's result [1]

Kopf's result [2]

Tan's result [3]

Fattal's result [4]

our result

input foggy image

He's result [1]

Kopf's result [2]

Tan's result [3]

Fattal's result [4]

our result


References:

[1] K.He, J. Sun, and X. Tang, "Single Image Haze Removal using Dark Channel Prior," in CVPR, 2009. (http://research.microsoft.com/en-us/um/people/kahe/cvpr09/index.html)

[2] J. Kopf, B. Neubert, B. Chen, M. Cohen, D. Cohen-Or, O. Deussen, M. Uyttendaele, and D. Lischinski, "Deep Photo: Model-Based Photograph Enhancement and Viewing," ACM Transactions on Graphics, 2008. (http://johanneskopf.de/publications/deep_photo/index.html)

[3] R. T. Tan, "Visibility in Bad Weather from a Single Image," in CVPR, 2008. (http://www.staff.science.uu.nl/~tan00109/fog)

[4] R. Fattal, "Single Image Dehazing," ACM Transactions on Graphics, 2008. (http://www.cs.huji.ac.il/~raananf/projects/defog)