Image shadow removal via multi-scale deep Retinex decomposition
Yan Huang,Xinchang Lu,Yuhui Quan,Yong Xu,Hui Ji
DOI: https://doi.org/10.1016/j.patcog.2024.111126
IF: 8
2024-11-08
Pattern Recognition
Abstract:In recent years, deep learning has emerged as an important tool for image shadow removal. However, existing methods often prioritize shadow detection and, in doing so, they oversimplify the lighting conditions of shadow regions. Furthermore, these methods neglect cues from the overall image lighting when re-lighting shadow areas, thereby failing to ensure global lighting consistency. To address these challenges in images captured under complex lighting conditions, this paper introduces a multi-scale network built on a Retinex decomposition model. The proposed approach effectively senses shadows with uneven lighting and re-light them, achieving greater consistency along shadow boundaries. Furthermore, for the design of network, we introduce several techniques for boosting shadow removal performance, including a shadow-aware channel attention module, local discriminative and Retinex decomposition loss functions, and a multi-scale mechanism for guiding Retinex decomposition by concurrently capturing both fine-grained details and large-scale contextual information. Experimental results demonstrate the superiority of our proposed method over existing solutions, particularly for images taken under complex lighting conditions.
computer science, artificial intelligence,engineering, electrical & electronic