An underwater imaging method of enhancement via multi-scale weighted fusion

Abstract

Blurring and color distortion are significant issues in underwater optical imaging, caused by light absorption and scattering impacts in the water medium. This hinders our ability to accurately perceive underwater imagery. Initially, we merge two images and enhance both the brightness and contrast of the secondary images. We also adjust their weights to ensure minimal effects on the image fusion process, particularly on edges, colors, and contrast. To avoid sharp weighting transitions leading to ghost images of low-frequency components, we then propose and use a multi-scale fusion method when reconstructing the images. This method effectively reduces scattering and blurring impacts of water, fixes color distortion, and improves underwater image contrast. The experimental results demonstrate that the image fusion method proposed in this paper effectively improves the fidelity of underwater images in terms of sharpness and color, outperforming the latest underwater imaging methods by comparison in PSNR, Gradient, Entropy, Chroma, AG, UCIQE and UIQM. Moreover, this method positively impacts our visual perception and enhances the quality of the underwater imagery presented.