Optimizing Color Consistency in Photo Collections

摘要

With dozens or even hundreds of photos in today’s digital photornalbums, editing an entire album can be a daunting task. Existingrnautomatic tools operate on individual photos without ensuring consistencyrnof appearance between photographs that share content. Inrnthis paper, we present a new method for consistent editing of photorncollections. Our method automatically enforces consistent appearancernof images that share content without any user input. Whenrnthe user does make changes to selected images, these changes automaticallyrnpropagate to other images in the collection, while stillrnmaintaining as much consistency as possible. This makes it possiblernto interactively adjust an entire photo album in a consistentrnmanner by manipulating only a few images.rnOur method operates by efficiently constructing a graph with edgesrnlinking photo pairs that share content. Consistent appearance ofrnconnected photos is achieved by globally optimizing a quadraticrncost function over the entire graph, treating user-specified edits asrnconstraints in the optimization. The optimization is fast enough tornprovide interactive visual feedback to the user. We demonstrate thernusefulness of our approach using a number of personal and professionalrnphoto collections, as well as internet collections.