水下
极化(电化学)
计算机科学
约束(计算机辅助设计)
直线(几何图形)
光学
地质学
海洋学
物理
几何学
数学
化学
物理化学
作者
Yiming Su,Zhen Zhang,Yang Lu,Shiben Liu,Yandong Tang,Jinshou Tian
标识
DOI:10.1016/j.optlaseng.2024.108265
摘要
It is effective to use polarization information for underwater descattering. However, existing methods always consider only the degree of linear polarization (DoLP) and ignore the angle of linear polarization (AoLP). In this paper, we propose a novel polarization imaging model on the complex plane that considers both the DoLP and the AoLP. Our model unifies and generalizes the polarization imaging model. Based on the model, we design an effective descattering method that can restore the intensity and polarization information from degraded images. In our method, we propose a polarization quantization to reduce polarization states from 1.8×1016 to only thousands. We propose a polarization line constraint, by which each distinct polarization state produces a polarization line in a degraded image on the complex plane. By the constraint, the clear image and its polarization states can be restored from its corresponding degraded image. To validate our method, we give an image quality metric for evaluating the accuracy of the restored polarization information. Experimental results show that our method is competitive with other state-of-the-art methods and achieves the best performance in terms of the underwater metrics.
科研通智能强力驱动
Strongly Powered by AbleSci AI