.png){kind=icon}
.png){kind=icon}
Return
Image harmonization in complex degradation scenes
.png){kind=icon}
PRE
AI
DOI：10.1016/j.patcog.2025.112227.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• The task of image harmonization in complex degradation scenes is proposed for the first time. • The degradation-aware conditional diffusion model is proposed to tackle the image harmonization problem. • The degradation map estimation module (DEM) is introduced to explicitly model degradations and the estimated degradation map is applied to guide diffusion process. • Masked deep supervision (MDS) is designed to optimize the proposed model and facilitates the generation of photorealistic harmonized images with reduced number sampling steps.
Keywords:
image harmonization
degradation-aware model
diffusion model
degradation map estimation
masked deep supervision
Journal
IF:
7.6
Papers:
1.3W
Citations:
4.5W