Image compositing aims to combine elements from multiple images producing natural-looking results. The Poisson-image-editing framework can generate seamless compositions, but tends to introduce color changes in the inserted elements, which may generate unrealistic and unpleasing results. Recently, Henz et al. have proposed an approach for controlling the amount of color changes in the inserted image patches, while ensuring a seamless transition between the foreground and background elements ([1]). We improve Henz et al.'s work by using new sets of basis vectors to span the color space, and by independently controlling the amount of adjustment in the individual color channels. This approach significantly enhances the freedom of the users to produce composites that were unreachable by the original technique. We have also performed a qualitative evaluation of the resulting technique, and show that it provides the flexibility to produce even more natural and pleasing composites when compared to previous techniques.

Keywords

Color adjustment, Color preservation, Poisson image editing, Seamless cloning, Image composition.

Paper

Pre-print Version (26 MB)




Pre-print Version (low resolution, 2.5 MB)

This is the author's version of the work. It is posted here by permission of Elsevier for your personal use. Not for redistribution. The final publication is available via http://dx.doi.org/10.1016/j.cag.2016.03.004.

Code

Code

Composition Results

Citation

Bernardo Henz, Frederico A. Limberger, Manuel M. Oliveira. Independent color-channel adjustment for seamless cloning based on Laplacian-membrane modulation, Computers & Graphics, Volume 57, June 2016, Pages 46-54, ISSN 0097-8493, http://dx.doi.org/10.1016/j.cag.2016.03.004.

BibTeX

@article{HenzLimbergerOliveira2016,
	title = "Independent color-channel adjustment for seamless cloning based on Laplacian-membrane modulation",
	journal = "Computers & Graphics",
	volume = "57",
	number = "",
	pages = "46 - 54",
	year = "2016",
	issn = "0097-8493",
	doi = "http://dx.doi.org/10.1016/j.cag.2016.03.004",
	url = "http://www.sciencedirect.com/science/article/pii/S0097849316300176",
	author = "Bernardo Henz and Frederico A. Limberger and Manuel M. Oliveira",
  }  
  

This work was sponsored by CAPES and CNPq-Brazil (grants 306196/2014-0, 134134/2013-3, 482271/2012-4, and 11892-13-7). The source image of Trolls is from http://www.alphamatting.com/. The remaining images are from Flickr users bernardohenz, Don McCullough, Bud, slworking, salemstudioz, simens, Wonderlane, didi8, cameliatwu, and Andrew Taylor; and from Pixabay user HebiFot. All the images are under Creative-Commons license.