An Analytical Review of Underwater Image Enhancement Techniques

Underwater images are integral parts for both terrestrial and aerial environments. Due to the environmental conditions in the deep-sea, poor visibility, haziness, and low contrast causes the distortion of the underwater images and hence image enhancement techniques are significant. Effects of color also plays a major role as refection, refraction, and scattering occurs. The water quality is controlled and influenced by the filtering properties of the water. Researchers proposed Underwater Image Enhancement (UIE) methods to solve these problems. In this paper, analysis has been done for five state of the art Underwater Image Enhancement models namely FUnIEGAN model, U-Shape Transformer model, DeepSeeColor model, Waternet model, and SRCNN model. To improve the quality of the image in terms of noise levels, Peak Signal-to-Noise Ratio (PSNR) is used and to compare images based on characteristics such as luminance, contrast, and structure, Structural Similarity Index (SSIM)is used. Results shows an analytical comparision among the enhancement models, and suggests one among them as superior based on the evaluation metrices.