Underwater optical imaging technology presents broad application prospects in fields such as marine resource exploration, underwater ecological environment monitoring, and seabed topography detection. The technology employs the polarization characteristics of light, particularly those of the background and target, to achieve a clear image. However, the traditional methods rely on target-free regions to compute the backscattered light information, which is infrequently present in the actual scene captured by the camera. Then the full-space resolution of target information light and backscattered light information are required. At this time, the traditional methods may be difficult to adapt in practical application.
In this paper, an underwater polarization de-scattering method independent of target-free regions is proposed by combining active polarization imaging and transmittance de-scattering model. Initially, the total light intensity within the camera's field of view is decomposed into its polarized and unpolarized components. By removing the backscattered light with polarized and unpolarized information from the total light intensity, a clear underwater target can be obtained. Based on the active polarization imaging model, the backscattered light with polarization information is calculated, in which the polarization angle of the backscattered light is considered as zero in the full-space. Thus, the polarization degree of the target information light occupying the camera's entire field of view can be derived. According to the polarization correlation, the polarization degree of the backscattered light can be characterized, and the intensity of the backscattered light with polarization information in the camera's entire field of view can also be obtained. Then the unpolarized component is calculated using the minimum intensity image with Stokes vector transformation. Finally, the underwater scene is obtained by combining the transmittance de-scattering principle and introducing adjustment parameters.
Experimental and real-world underwater imaging results demonstrate that the proposed method can effectively remove the majority of the backscattered light and improve the image contrast and entropy, irrespective of the presence of target-free regions. Additionally, it offers a certain rate advantage, which can facilitate the real-time complex underwater imaging technology.