10
16

Deep Image Harmonization in Dual Color Spaces

Abstract

Image harmonization is an essential step in image composition that adjusts the appearance of composite foreground to address the inconsistency between foreground and background. Existing methods primarily operate in correlated RGBRGB color space, leading to entangled features and limited representation ability. In contrast, decorrelated color space (e.g., LabLab) has decorrelated channels that provide disentangled color and illumination statistics. In this paper, we explore image harmonization in dual color spaces, which supplements entangled RGBRGB features with disentangled LL, aa, bb features to alleviate the workload in harmonization process. The network comprises a RGBRGB harmonization backbone, an LabLab encoding module, and an LabLab control module. The backbone is a U-Net network translating composite image to harmonized image. Three encoders in LabLab encoding module extract three control codes independently from LL, aa, bb channels, which are used to manipulate the decoder features in harmonization backbone via LabLab control module. Our code and model are available at \href{https://github.com/bcmi/DucoNet-Image-Harmonization}{https://github.com/bcmi/DucoNet-Image-Harmonization}.

View on arXiv
Comments on this paper