| Underwater optical imaging technology is an important means to detect underwater targets,but because the light transmitted underwater will be affected by water absorption and scattering,the contrast and clarity of underwater target images obtained using traditional intensity imaging are low.The target information contained is also less,so how to obtain high-quality underwater target images is an important issue.Polarization imaging can effectively reduce the influence of underwater scattered light.The polarization image obtained by this method can better reflect the contour edge information and texture characteristics of the target,so the target information contained in the image is also richer.In order to improve the image quality,further image processing of polarization degree image is needed.n this paper,carbon fiber tubes,beverage cans,adhesive tapes and rubber dolls are used as underwater imaging targets,and the detection and image processing methods of underwater target laser active polarization imaging are studied.The main research contents and work are as follows:1.A single-channel rotating time-sharing laser active polarization imaging system is built to obtain the polarization degree image of the underwater target.Aiming at the problem that the polarization degree image is easily affected by noise,the three filtering algorithms,namely median filter,mean filter and improved homomorphic filter,are analyzed and compared in their ability to remove the noise of polarization degree image.The obtained polarization degree image processing works best.2.The intensity image and the polarization image were compared and analyzed through experiments.The experimental results show that in the underwater environment,the average edge intensity value of the polarization image is 6.3374 times that of the intensity image,indicating that the polarization image has better edge characteristics and can better highlight the target in the underwater environment.3.An image fusion method based on Retnix and wavelet transform is proposed.The high and low frequency components of the image are processed by the fusion rules of taking large pixel value and saliency map respectively.The experimental results of image fusion show that compared with the five commonly used image fusion methods,the fusion image obtained by this method has better visual effect,higher definition and more obvious image details.The image has an average improvement of 21.63%,53.20%,and 51.48% in information entropy,contrast,and average gradient. |