基于生成对抗网络的轻量级全局-局部水下图像增强算法

Lightweight global⁃local underwater image enhancement algorithm based on generative adversarial network

为解决光在水下传播过程中由吸收与散射效应导致的水下图像产生色偏、对比度低、细节缺失和噪音等问题,设计了一种融入注意力机制的轻量级全局-局部生成对抗网络水下图像增强算法。算法通过改进GAN网络以及设计新的损失函数,提高图像增强效果。实验结果表明,所提出的方法在校正色偏、提高对比度、增强细节和消除噪音等方面较现有流行的水下图像增强算法均取得了很大的进步,同时也进行了时间测试以及应用测试,实验结果表明,提出的水下图像增强方法能够快速有效地提高水下图像的质量。

In order to solve the problems of underwater imagecolor cast,low contrast,detail missing and noise caused by the absorption and scattering effects of light during underwater propagation,a generative confrontation network based lightweight global⁃local underwater image enhancement algorithm incorporating attention mechanism is designed.The algorithm can improve the image enhancement effect by perfecting the GAN network and designing a new loss function.The experimental results show that the proposed method has made great progress in correcting color cast,improving contrast,enhancing details,eliminating noise,etc.,compared with the existing popular underwater image enhancement algorithms.At the same time,time test and application test have also been carried out.The experimental results show that the proposed underwater image enhancement method can quickly and effectively improve the quality of underwater images.