Underwater Image Enhancement Based on Red Channel Correction and Improved Multiscale Fusion

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.