Underwater Targets Detection Based On Polarization Imaging

Underwater optical imaging technology plays an important role in underwater detection,including exploring marine resources,ocean rescue,aquaculture and so on.However,the absorption and scattering of light in water are two major reasons leading to image degradation,for example,reduction of intensity,loss of contrast,blurred details,et.al.Therefore,it's necessary to research methods to obtain high-quality underwater images,and further to enhance underwater detection ability.To meet the demands of high-quality underwater imaging,this dissertation focused on researching active underwater polarization imaging.It's possible to separate target irradiance and background irradiance by analyzing the change of polarization characteristics of light propagating underwater.This will help to detect and recognize underwater targets.All the optical imaging methods are based on the characteristics of light in water.And this study also started by modeling the propagation of light in water,including absorption model and scattering model.Analyses and results demonstrated that both the forward and backward scattering light can maintain polarization characteristics of light to some extent,but scattering of light depends on the turbidity of water.These facts make it possible to detect underwater targets employing polarization characteristics of light.By first researching how light is scattered in water,it's proved that polarization characteristics of light will help to suppress scattering light.However,polarization imaging and process generally lead to noise amplification in the final image.The amplification level is proportional to the number of captured polarization images.Based on this fact,an active polarization imaging method based on optical correlation is proposed to address the noise amplification issue.By determining two best polarization azimuthal images,the introduced noise can be minimized,which further provides improved underwater images.Besides,the final detection results are predictable based on the captured polarization azimuthal images.Forward scattering generally leads to blurred images by randomly deviating light on itsway from the object to the camera.The backward scattering veils the image and reduces the image contrast by reflecting the light to the camera before the light reaches the target object.The human visual system is typically sensitive to the contrast variation,therefore,traditional methods focused on addressing backscattering light.However,neglect of forward scattering may lead to target missing since forward scattering is common in different natural water and it grows with Mie scattering growing.This study attempts to deal with the image degradation caused by forward scattering based on the image degradation model.Taking the image degradation as the result of backward scattering and forward scattering,the forward scattering light can be eliminated by estimating its degradation function and reconstructing a clear image according to the degradation function.Positive results are achieved by experiments and simulations.Till now,our research are majorly completed for water with a rather low concentration of floating particles.When water gets rather turbid,it's hardly possible to obtain effective target information by traditional polarization imaging methods.In this situation,we analyzed the scattering and absorption characteristics of turbid water,and proposed an active polarization imaging method with illumination at a specially selected wavelength in turbid water.By first establishing a polarization imaging model in strong scattering water,clear images are finally achievable by estimating the degree of polarization of target irradiance and background irradiance and eliminating background irradiance from images.Experiments demonstrated the feasibility of the proposed method in imaging through turbid water.It's potential for underwater rescue in river,lake or coastal water where water is highly turbid.