Color Correction Method using Monocular Depth Estimation Model for Underwater Images

In this work, a new color correction algorithm is proposed to reduce degradation in underwater images. The color correction algorithm depends on the generation of high qulity depth map. In conventional methods, the depth map is generated by Monodepth2. However, Monodepth2 is a deep learning-based method that requires the training time, making it difficult to adapt to changes in the environment. Therefore, our proposed algorithm uses Udepth, a physics-based method that doesn’t use deep learning. The Udepth method generates the depth map only on the basis of the RGB values. It could quickly respond to changes in the underwater environment by adjusting high parameters. As a result, with the high accuracy depth map the color correction is improved.