Water-Filling: An Efficient Algorithm for Digitized Document Shadow Removal

Original.

Bako et al.

Kligler et al.

Sternberg et al.

Tableau et al.

Zhang et al.

Water-Filling.

GroundTruth

My second year project under advisor Prof. Changick Kim (2015.09~2016.03). The technique is for removing illumination distortions for camera captured document images. Project funded by Hancom company.

Related papers is accepted: Seungjun Jung, Muhammad Abul Hasan, and Changick Kim, "Water-Filling: An Efficient Algorithm for Digitized Document Shadow Removal," ACCV, December 2018.

Benchmark dataset: Supplemental materials

Project Funded by Hancom Inc.

Requirements

opencv 3.0
c++

Thanks

Thank you Yun-Hsuan Lin, and Jian-Ren Wang for citing my work!

In 2016 ACCV, Bako, and I talked that this field only gets a few public attention. Now I am glad that many people are showing interest on this (and on Bako's) work.

About me

I am not currently working on this field... :D

Note

Source Code Available, but do not use it (for industrial purpose) without permission. (Patent Issue ... )
Please ask any question if you have : [email protected]
Sample Results are also Available.

Tips

(2020/04/22) For those who are interested in applying this method in printed colored documents (i.e., magazine), "alexliyang" suggests that spliting image channels would help improving performance. (Thank you!)

If you think the results are too bright, you can adjust brightness factor from 0.85 to lower! :)