Computers & Graphics 83, 42–50. 2019.

Polarization mapping

Fernando del Molino Adolfo Muñoz¹

¹ Universidad de Zaragoza, I3A

Polarization-based photographic editing. (a) Original image, without any polarizing filter. (b) The same scene, filtered with a linear polarizer automaticaly rotated (optimized) for minimizing the saturation in the image. (c) The same scene, applying a different linear polarizer per pixel to maximize each pixel’s luminance. The optimal angle for luminance maximization can be different at each pixel. (d) The same scene edited with our provided brush based tool, that illustrates that all filters work at interactive rates. In the building, the luminance of the left part is minimized and on the right part is maximized. In the sky, the right area has a maximized saturation, while the left are has minimized saturation, all of it with linear per-pixel polarizing filters.

Abstract

Photographers use the hardware of the camera (aperture, exposure time…), lens and filters as tools for artistic expressivity. This expressivity has often been enhanced by software, such as high dynamic range images have been edited in post-process with software tone mappers. In this paper, we propose a similar approach with polarization filters: we design a capture process that enables us to acquire a Stokes image (that encodes all the possible light polarization states) with a single camera, and we then offer a set of software tools that can apply any common polarization filter as a software post-process, delaying the choice of the adequate filter and enabling filters that can be mathematically modeled but are not available as hardware. Then, we devise and provide new algorithms that automatically select the optimal filter for specific goals, such as maximizing (or minimizing) brightness, contrast or saturation. We later show how such optimization filters can not only be applied to the whole image, but can also be at per-pixel level, obtaining new interesting effects. Such optimization can work at real time rates, fact that is illustrated with a brush based user editing interactive tool. The different types of filters are tested in a wide range of results.

Files

Source code

The source code for this paper has been released to the public domain.

GitHub.

News

This paper was published through the Computers & Graphics track of the Spanish Conference on Computer Graphics 2019 (CEIG2019).
This paper was selected as best paper of the Spanish Conference on Computer Graphics 2019 (CEIG2019).

Video

BibTeX

Acknowledgements

We thank the anonymous reviewers for their useful suggestions and insights. This project has been funded by DARPA (project REVEAL), the European Research Council (ERC) under the EU’s Horizon 2020 research and innovation programme (project CHAMELEON, grant No 682080) and the Spanish Ministry of Economy and Competitiveness (project TIN2016-78753-P).