Restoration of underwater vision using a two-phase regularization mechanism

Abstract

Clear underwater vision is crucial to many research and realistic applications in ocean engineering, ocean biology, and ocean science. However, owing to particular propagation properties of light beneath water, underwater images usually have poor quality including low contrast, blurring, darkness, and color diminishing. In this article, a new underwater image restoration algorithm is proposed that consists of two major phases: haze removal and color correction. In the first phase, underwater images are observed similar to haze images because they have the same problems of low contrast and color shifting. This motivated us to use the dark channel prior haze removal technique to dehaze underwater images. Subsequently, in the second phase, we equalize the color mean in each RGB (red, green, blue) channel to balance the color. We use the CLAHE method to further increase the local contrast of the image. Then, the color space is transformed from RGB to HSV (hue, saturation, value) to adjust the S channel to make the image color more natural. Finally, we adjust V channel according to the brightness value of RGB to enhance the global contrast. Preliminary results indicated that the proposed method effectively improved visual interpretation of underwater images.