A method for quickly recoloring photos by extracting dominant colors for editing, with masks for selective changes and preserved image quality. It runs 3 to 20 times faster than Photoshop.
Dear fellow scholars, this is two minute Papers with Image and Color editing is an actively researched topic with really cool applications that you will see in a second. Most of the existing solutions are either easy to use but lack in expressiveness, or they are expressive but too complex for novices to use. Computation time is also an issue as some of the operations in Photoshop can take more than a minute to carry out. For color editing, the workflow is very simple. The program extracts the dominant colors of an image which we can interactively edit ourselves. An example use case would be recoloring the girl's blue sweater to turquoise or changing the overall tone of the image to orange. Existing tools that can do this are usually either too slow or only accessible to adept users. It is also important to note that it is quite easy to take these great results for granted. Using a naive color transfer technique would destroy a sizable part of the dynamic range of the image and hence legitimate features which are all preserved. If we use this algorithm instead, one can also use masks to selectively edit different parts of the image. The technique executes really quickly, opening up the possibility of not real time but interactive recoloring of animated sequences. Or you can also leverage the efficiency of the method to edit not one, but a collection of images in one go. The paper also contains a rigorous evaluation against existing techniques. For instance, they show that this method executed 3 to 20 times faster than the one implemented in Adobe Photoshop. Thanks for watching and for your generous support and I'll see you next time.
