Abstract:In order to solve the problems about color distortion and low contrast of underwater images, we propose an underwater image enhancement algorithm that combines adaptive color correction with improved Retinex algorithm. Our algorithm is a single-image enhancement method that does not require specialized hardware and underwater scenes prior. Firstly, the adaptive color correction is carried out on the underwater distorted images to solve the color cast problem effectively. Then, on the one hand, we use the image decomposition to strengthen the detail part and obtain a detail enhanced image. On the other hand, we use the improved Retinex algorithm to strengthen the edge part and obtain an edge enhanced image. Finally, the detail enhanced image and the edge enhanced image are fused based on the non-subsampled shearlet transform (NSST) to obtain the final enhanced underwater image. The results show that our method outperforms several state-of-the-art methods about underwater image enhancement in terms of PCQI, UCIQE, UIQM and IE. By scale invariant feature transform (SIFT) algorithm, we calculate the number of feature matching points of the input image and the enhanced image, and our proposed method achieves the best experimental results. The source code of our proposed algorithm is available at: https://github.com/lin9393/ underwater-image-enhance.
computer science, information systems,telecommunications,engineering, electrical & electronic
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the color distortion and low contrast of underwater images. Specifically, the author proposes an underwater image enhancement method that combines adaptive color correction with an improved Retinex algorithm, aiming to improve the visual effect of underwater images, making improvements in terms of color fidelity, detail clarity, and overall contrast. This method does not rely on dedicated hardware or prior knowledge of underwater scenes and is suitable for the enhancement processing of a single image.
### Background of the Paper and Problem Description
With the increasing shortage of terrestrial resources, the ocean has become the focus of exploration because of its rich oil and mineral resources. Obtaining high - quality and clear underwater images is an important guarantee for the development of underwater vehicle control, underwater infrastructure inspection, and marine biology research. However, due to the attenuation and scattering of light in water, underwater images often have problems such as color deviation, image blurring, and low contrast. In order to promote subsequent research and applications, it is particularly important to improve the visual effect of underwater images.
### Overview of the Solution
The method proposed in the paper mainly includes the following steps:
1. **Adaptive Color Correction**:
- Solve the color shift problem of underwater images through the adaptive color correction method. This method first compensates for the red, green, and blue channels, and then performs a color - balance operation to adjust the overall hue of the image.
2. **Two - scale Image Decomposition and Gamma Correction**:
- Use large - scale mean - filtering to decompose the color - corrected image into a base layer and a detail layer, adjust the detail layer through an enhancement coefficient, and then perform Gamma correction to enhance the local brightness and contrast of the image.
3. **Improved Retinex Algorithm**:
- Adjust the hue of the image through the improved Retinex algorithm, especially enhancing the edge parts of the image. This algorithm includes three parts: global adaptation, local adaptation, and color balance, and can effectively deal with the problem of detail loss in high - brightness areas.
4. **Image Fusion**:
- Finally, use the non - subsampled shearlet transform (NSST) technique to fuse the detail - enhanced image and the edge - enhanced image to obtain the final enhanced underwater image.
### Experimental Results and Evaluation
The paper evaluates the proposed algorithm from both subjective and objective aspects. The subjective comparison results show that the algorithm performs well in different types of underwater images (such as blue, green, yellow, and blurred images), and can effectively solve problems such as color deviation, noise elimination, and detail preservation. Objective evaluation indicators include PCQI, UCIQE, UIQM, and IE. The experimental results show that the algorithm is superior to several existing advanced methods in these four indicators.
### Conclusion
This paper proposes an underwater image enhancement method that combines adaptive color correction with an improved Retinex algorithm, which can effectively solve the problems of color distortion and low contrast of underwater images and improve the visual effect of underwater images. The performance of this method on multiple data sets is superior to existing methods and has high practical value.