多尺度扩散模型用于水下图像恢复和增强.
Yuan Fang1, Qianyi Li2, Kai Wang3
1School of Innovation and Entrepreneurship, Dalian Polytechnic University, Dalian, Liaoning, China.
PloS one
|September 10, 2025
概括
这项研究介绍了一种新的水下图像增强算法,它结合了物理模型和深度学习. 该方法显著提高了图像质量,推进了水下勘探和监测能力.
科学领域:
- 计算机视觉 计算机视觉
- 图像处理 图像处理
- 深度学习 (Deep Learning) 是一种深度学习.
背景情况:
- 水下图像因光吸收和散射而受损,导致模糊,低对比度和色彩扭曲.
- 这种退化阻碍了水下勘探,环境监测和科学研究.
- 有效的图像增强算法对于克服这些挑战至关重要.
研究的目的:
- 开发一种创新的水下图像增强算法.
- 将物理模型与深度学习相结合,以获得卓越的视觉质量.
- 在关键绩效指标上超越现有方法.
主要方法:
- 一个多层次的设计,将物理扩散模型的见解与深度学习相结合.
- 一个编码器-解码器管道用于图像分解和自适应色彩调整.
- 一个用于像素级优化的反向无光扩散模型.
主要成果:
- 在峰值信号与噪声比率 (PSNR),结构相似性指数测量 (SSIM) 和水下图像质量测量 (UIQM) 中观察到显著改善.
- 在PSNR中平均增加了6.61,在SSIM中增加了0.15,在UIQM中增加了0.87.
- 物理模型和深度学习的整合证明了有效性.
结论:
- 拟议的算法为增强水下图像提供了一个强大的框架.
- 这项研究推动了水下图像处理领域的发展.
- 这些发现支持改善用于勘探和监测的视觉数据.
相关概念视频
Modeling and Similitude
620
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
620
Typical Model Studies
620
Fluid mechanics model studies often utilize scaled-down systems to predict fluid behavior in full-scale environments, such as river flows, dam spillways, and structures interacting with open surfaces. Maintaining Froude number similarity in river models is crucial, as it replicates surface flow features like wave patterns and velocities.
620


