Underwater image enhancement by modified color correction and adaptive Look-Up-Table with edge-preserving filter

Abstract

Particles suspended in water cause poor visibility and color distortions in the images that are captured in underwater. These suspended particles cause light to be scattered that results in non-uniform illumination due to dispersion and refraction of light. The illumination variation cause color casts in the underwater images. This paper proposes an underwater image enhancement method that improves the visibility, contrast and color cast in the underwater images. The proposed method includes a modified color correction, an adaptive Look-Up-Table (LUT) and an edge-preserving filter. First, a color correction is performed on individual LAB color space components. The improvement of the luminance (L) component is processed with a sequence of steps such as CLAHE, sharpening, gamma and color correction. In addition, the chrominance (A and B) components are modified by further color correction. Second, the contrast of the color-corrected image is enhanced using the adaptive LUT based on probability threshold. This probability threshold increases the image contrast by limiting the pixel intensity values in each RGB channel. In addition, the probability threshold controls the halo effects and artifacts. Finally, the underwater image is enhanced with a fast local Laplacian filter (FLLF) to preserve the edge details with texture smoothing. The denoised image is adjusted with histogram normalization for a further color balancing. The experimental results obtained from the proposed method on different underwater image datasets shows a supremacy over color casts, visibility, halo effects, free of artifacts and edge-preserving in the underwater images compared to the existing methods.