An Enhanced Multi-Stage Approach for Dehazing Underwater Images

Tamilarasi Kathirvel Murugan,Shivansh Sharma,Agniv Ganguly,Ashesh Banerjee,Keshav Kejriwal

DOI: https://doi.org/10.1109/access.2024.3486456

IF: 3.9

2024-11-01

IEEE Access

Abstract:Suffering of underwater images is often a multitude of issues, including low contrast, detail blurring, color attenuation, and poor illumination, primarily caused by light scattering and absorption in water. In response to these challenges, the study introduces a two-step method aimed at addressing these issues. The method involves the utilization of a "zero-shot" dehazing network followed by enhanced adjustment of level technique with auto contrast combination. Initially, the original image undergoes processing using the "zero-shot" dehazing network, after which it is further refined through an improved level adjustment methodology. To assess its effectiveness, comparative experiments were carried out to evaluate the proposed methodology's performance against the efficient six classical state-of-the-art methods. The proposed methodology reveals the successful elimination of haze, corrects color deviations, and preserves the natural appearance of the images. Furthermore, the quantitatively evaluated proposed methodology surpasses the peak signal-to-noise ratio and structural similarity based on compared methods. This indicates its superior performance in restoring image quality. The efficiency of the proposed methodology of enhanced underwater blurred images is validated by the experimental results, highlighting its potential for practical application in addressing the challenges associated with underwater image processing.

computer science, information systems,telecommunications,engineering, electrical & electronic