Deep Image Harmonization
This addresses a common issue in photo editing for users who create composites, offering a more reliable solution than hand-crafted feature methods, though it is incremental as it builds on deep learning approaches.
The paper tackles the problem of making composite images look realistic by harmonizing foreground and background appearances, proposing an end-to-end deep convolutional neural network that outperforms previous state-of-the-art methods on both synthesized and real composite images.
Compositing is one of the most common operations in photo editing. To generate realistic composites, the appearances of foreground and background need to be adjusted to make them compatible. Previous approaches to harmonize composites have focused on learning statistical relationships between hand-crafted appearance features of the foreground and background, which is unreliable especially when the contents in the two layers are vastly different. In this work, we propose an end-to-end deep convolutional neural network for image harmonization, which can capture both the context and semantic information of the composite images during harmonization. We also introduce an efficient way to collect large-scale and high-quality training data that can facilitate the training process. Experiments on the synthesized dataset and real composite images show that the proposed network outperforms previous state-of-the-art methods.