基于颜色注意机制的弱光图像增强算法

Low-Light Image Enhancement Algorithm Based on Color Attention Mechanism

针对当前弱光图像增强算法在恢复过程中存在颜色与细节丢失的问题,提出一种基于颜色注意机制的增强算法.该算法首先将弱光图像从RGB色彩空间转换至CIE LAB色彩空间,将弱光图像分解为亮度和颜色两个分量.其次,利用两个卷积神经网络(CNN)模型对亮度和颜色分量分别独立增强.然后对增强后的颜色分量使用监督注意力机制,在弱光图像中搜索任何有用的颜色关键点,引导和扩展网络注意力对图像的颜色进行增强.最后将增强后的亮度和颜色分量进行融合,并转换回RGB空间,获得清晰艳丽的复原图像.实验结果表明:与其他增强算法相比,本文提出的方法具有明显优势,在保持明暗度顺序的同时有效地增强了弱光图像,完整地恢复了原始图像的颜色.

Aiming at the problem of loss of color and details in the restoration process of the current low-light image enhancement algorithms,an enhancement algorithm based on the color attention mechanism has been proposed.Firstly,the algorithm converts the low-light image from the RGB color space to the CIE LAB color space,and decomposes the low-light image into brightness component and color component.Secondly,the two convolutional neural network(CNN)models are used to enhance the brightness and color components independently.Thirdly,the algorithm uses the supervised attention mechanism to search for any useful color key points in the low light image,and guides and expands the attention of the network to enhance the color of the image.And finally,the enhanced brightness and color components are fused and converted back to RGB space to obtain a clear and gorgeous restored image.Experimental results show that compared with other enhancement algorithms,the proposed method has obvious advantages.It effectively enhances the low-light image while maintaining the order of lightness and darkness,and faithfully restores the original image color.