The underwater environment is complex and ever-changing, and color distortion and reduced contrast are the two primary issues encountered in original underwater images. This article applies image enhancement to address these concerns by proposing a novel image color correction red channel correction after gray world (RAG) algorithm and an improved multiscale image fusion (IMF) algorithm for each problem. In terms of color correction, traditional algorithms typically implement a single color correction step, leading to issues such as locally unrealistic colors and limited applicability in certain scenes. The proposed RAG algorithm advances upon these methods by further dividing the input image into multiple local blocks after performing global color correction using the gray world algorithm. Secondary color corrections are then applied within each local block, overcoming the limitations of the traditional single-step color correction process. The corrected images achieve an underwater image colorfulness measure (UICM) mean value of 5.4671, representing an enhancement exceeding 20%. In contrast, the existing fusion algorithms often neglect the distinct characteristics of different channels and fail to maximize the advantages of fusion. The IMF algorithm proposed in this article calculates weights in the red green blue color space (RGB) and LAB color spaces of images to effectively incorporate channel disparities, refining the weights to provide a more thorough fusion process. The corrected images achieve a mean underwater color image quality evaluation (UCIQE) value of 0.630 and a mean colorfulness, contrast and fogdensity (CCF) value of 37.495. Significant improvements are observed across various other metrics.

中文翻译：

---

基于红色通道校正和改进多尺度融合的水下图像增强

水下环境复杂多变，色彩失真和对比度降低是原始水下图像遇到的两个主要问题。本文应用图像增强来解决这些问题，针对每个问题提出了一种新颖的图像颜色校正后灰色世界校正红色通道校正（RAG）算法和改进的多尺度图像融合（IMF）算法。在色彩校正方面，传统算法通常实现单一的色彩校正步骤，从而导致色彩局部不真实、在某些场景下适用性有限等问题。所提出的 RAG 算法在这些方法的基础上进行了改进，在使用灰色世界算法执行全局颜色校正后，进一步将输入图像划分为多个局部块。然后在每个局部块中应用二次颜色校正，克服了传统单步颜色校正过程的局限性。校正后的图像的水下图像色彩度测量 (UICM) 平均值为 5.4671，增强超过 20%。相比之下，现有的融合算法往往忽略了不同通道的独特特征，无法最大限度地发挥融合的优势。本文提出的IMF算法计算图像的红绿蓝颜色空间（RGB）和LAB颜色空间中的权重，以有效地融合通道差异，细化权重以提供更彻底的融合过程。校正后的图像的平均水下彩色图像质量评估 (UCIQE) 值为 0.630，平均色彩度、对比度和雾密度 (CCF) 值为 37.495。在其他各种指标上都观察到了显着的改进。