Enhancement of Underwater Images with Statistical Model of Background Light and Optimization of Transmission Map (1906.08673v1)

Abstract: Underwater images often have severe quality degradation and distortion due to light absorption and scattering in the water medium. A hazed image formation model is widely used to restore the image quality. It depends on two optical parameters: the background light and the transmission map. Underwater images can also be enhanced by color and contrast correction from the perspective of image processing. In this paper, we propose an effective underwater image enhancement method for underwater images in composition of underwater image restoration and color correction. Firstly, a manually annotated background lights (MABLs) database is developed. With reference to the relationship between MABLs and the histogram distributions of various underwater images, robust statistical models of BLs estimation are provided. Next, the TM of R channel is roughly estimated based on the new underwater dark channel prior via the statistic of clear and high resolution underwater images, then a scene depth map based on the underwater light attenuation prior and an adjusted reversed saturation map are applied to compensate and modify the coarse TM of R channel. Next, TMs of G-B channels are estimated based on the difference of attenuation ratios between R channel and G-B channels. Finally, to improve the color and contrast of the restored image with a natural appearance, a variation of white balance is introduced as post-processing. In order to guide the priority of underwater image enhancement, sufficient evaluations are conducted to discuss the impacts of the key parameters including BL and TM, and the importance of the color correction. Comparisons with other state-of-the-art methods demonstrate that our proposed underwater image enhancement method can achieve higher accuracy of estimated BLs, less computation time, more superior performance, and more valuable information retention.

• The paper introduces a novel framework integrating statistical background light estimation and multi-step transmission map refinement to enhance image clarity.
• It establishes a manually annotated underwater image database (MABLs) that leverages histogram distributions for robust background light modeling.
• The method outperforms existing techniques by significantly lowering RMSE and boosting SSIM while reducing computational costs.

Enhancement of Underwater Images with Statistical Model of Background Light and Optimization of Transmission Map

The research paper by Wei Song et al. presents a comprehensive and effective methodology for underwater image enhancement. Underwater imaging is crucial for various applications, including underwater surveillance and marine life paper, where image quality is significantly degraded due to light absorption and scattering in the water medium. The paper details a novel technique that integrates underwater image restoration and color correction to achieve enhanced image quality.

The authors introduce a manually annotated background lights (MABLs) database, comprising 500 underwater images, to support their statistical model for background light (BL) estimation. This model utilizes the histogram distributions of underwater images and demonstrates improved accuracy and computational efficiency over existing BL estimation methods.

Transmission maps (TM), essential for estimating the portion of scene radiance reaching the camera without scattering, are refined through a New Underwater Dark Channel Prior (NUDCP), a scene depth map derived from an underwater light attenuation prior (ULAP), and an adjusted reversed saturation map (ARSM). This multi-step process corrects coarse TM estimates and adapts them to the unique properties of the underwater environment, thus enhancing the fidelity of the restored images.

The authors further employ a white balance variation to improve color and contrast, ensuring that the restored underwater images exhibit a more natural appearance. A comparison of their method against various existing approaches, such as DCP, UDCP, and Peng's method, highlights superior performance in terms of noise reduction, structural similarity, and retention of valuable image information, at notably reduced computational cost.

Key numerical results include significant improvements in standard image quality metrics, such as a lower Root Mean Square Error (RMSE) and higher Structural Similarity Index (SSIM). In essence, this paper provides a robust framework for underwater image enhancement that holds promise for practical applications requiring high-quality visual data.

The implications extend beyond immediate practical enhancement; the establishment of the MABLs database marks an important resource for ongoing research into image restoration. The authors advocate for future growth of this database to enhance model robustness, potentially employing machine learning techniques to explore further improvements in BL and TM estimation.

In conclusion, this paper advances the domain of underwater image processing and sets a foundation for further research into adaptive image enhancement techniques, which might be leveraged in other domains of image processing and computer vision.