ShadowHack: Hacking Shadows via Luminance-Color Divide and Conquer

Jin Hu,Mingjia Li,Xiaojie Guo

2024-12-04

Abstract:Shadows introduce challenges such as reduced brightness, texture deterioration, and color distortion in images, complicating a holistic solution. This study presents \textbf{ShadowHack}, a divide-and-conquer strategy that tackles these complexities by decomposing the original task into luminance recovery and color remedy. To brighten shadow regions and repair the corrupted textures in the luminance space, we customize LRNet, a U-shaped network with a rectified outreach attention module, to enhance information interaction and recalibrate contaminated attention maps. With luminance recovered, CRNet then leverages cross-attention mechanisms to revive vibrant colors, producing visually compelling results. Extensive experiments on multiple datasets are conducted to demonstrate the superiority of ShadowHack over existing state-of-the-art solutions both quantitatively and qualitatively, highlighting the effectiveness of our design. Our code will be made publicly available at <a class="link-external link-https" href="https://github.com/lime-j/ShadowHack" rel="external noopener nofollow">this https URL</a>

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the complex degradation problems in shadow removal, which include the reduced brightness in shadow areas, texture degradation and color distortion. Shadow removal is a challenging task because shadows not only lead to the reduction of brightness in certain areas of the image, but also cause the loss of texture information and color deviation. These factors are intertwined, making a single overall solution difficult to achieve. Although the existing methods have improved the effect of shadow removal to a certain extent, they still have deficiencies in restoring color and underlying texture.

To solve the above problems, this paper proposes the ShadowHack framework, which independently deals with these problems by decomposing the original task into two subtasks: brightness restoration and color correction. Specifically, ShadowHack first uses a simple but very effective shadow decomposition technique to decompose the input image into brightness and color components. Then, two dedicated networks are established respectively - the Brightness Restoration Network (LRNet) and the Color Regeneration Network (CRNet). Among them, LRNet focuses on restoring brightness and texture, while CRNet uses the cross - attention mechanism to query accurate color information from non - shadow areas to restore the color of shadow areas. Through this divide - and - conquer strategy, ShadowHack can remove shadows more effectively and provide more visually appealing results.

In addition, ShadowHack introduces an innovative mechanism - Rectified Outreach Attention (ROA). This mechanism realizes the accurate restoration of contaminated textures and irregular illumination by expanding the receptive field of the window - based attention module and recalibrating the noise. This mechanism improves the repair quality of shadow areas while maintaining a low computational cost.

In general, by decomposing the shadow removal task and designing dedicated networks for each subtask, ShadowHack successfully solves the complex degradation problems in shadow removal and provides better performance than the existing methods.