OceanLens: An Adaptive Backscatter and Edge Correction using Deep Learning Model for Enhanced Underwater Imaging

Rajini Makam,Dhatri Shankari T M,Sharanya Patil,Suresh Sundram

2024-11-20

Abstract:Underwater environments pose significant challenges due to the selective absorption and scattering of light by water, which affects image clarity, contrast, and color fidelity. To overcome these, we introduce OceanLens, a method that models underwater image physics-encompassing both backscatter and attenuation-using neural networks. Our model incorporates adaptive backscatter and edge correction losses, specifically Sobel and LoG losses, to manage image variance and luminance, resulting in clearer and more accurate outputs. Additionally, we demonstrate the relevance of pre-trained monocular depth estimation models for generating underwater depth maps. Our evaluation compares the performance of various loss functions against state-of-the-art methods using the SeeThru dataset, revealing significant improvements. Specifically, we observe an average of 65% reduction in Grayscale Patch Mean Angular Error (GPMAE) and a 60% increase in the Underwater Image Quality Metric (UIQM) compared to the SeeThru and DeepSeeColor methods. Further, the results were improved with additional convolution layers that capture subtle image details more effectively with OceanLens. This architecture is validated on the UIEB dataset, with model performance assessed using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM) metrics. OceanLens with multiple convolutional layers achieves up to 12-15% improvement in the SSIM.

Image and Video Processing

What problem does this paper attempt to address?

### Problems the paper attempts to solve

The paper aims to solve the problems of decreased image clarity, contrast, and color fidelity in underwater imaging due to the selective absorption and scattering of light in water. Specifically, the paper proposes a method named OceanLens, which models the physical characteristics of underwater images through neural networks, including backscattering and attenuation, and introduces adaptive backscattering and edge - correction loss functions to improve image quality.

### Main contributions

1. **Modeling the physical characteristics of underwater images**:
- **Backscattering**: Estimate the backscattering component through neural networks to reduce the scattering effect caused by suspended particles and microorganisms.
- **Attenuation**: Calculate attenuation correction through depth maps and direct images to restore the clarity and details of the original image.

2. **Adaptive loss functions**:
- **Sobel and LoG losses**: Used to manage image variance and brightness, enhancing edge detection and detail capture.
- **Adaptive Huber loss**: Combines robustness and smoothness to adapt to different lighting conditions.

3. **Monocular depth estimation model**:
- Utilize a pre - trained monocular depth estimation model to generate underwater depth maps, improving the effect of image enhancement.

4. **Performance evaluation**:
- Use the SeeThru dataset and the UIEB dataset for evaluation. The results show that OceanLens significantly outperforms existing methods in terms of Grayscale Patch Mean Angular Error (GPMAE) and Underwater Image Quality Metric (UIQM).
- On the UIEB dataset, OceanLens with multiple convolutional layers has achieved a 12 - 15% improvement in Structural Similarity Index Measure (SSIM).

### Conclusion

OceanLens effectively addresses the challenges in underwater image enhancement by combining neural network architectures and specialized loss functions. This method not only performs well in terms of image clarity, contrast, and color fidelity, but also verifies its superiority on multiple datasets, providing a new solution for underwater image processing.