Research On Underwater Three-dimensional Imaging And Image Processing Of Streak Tube Imaging Lidar

With the development of miniaturized high-repetition rate,high-power laser technology and new optoelectronic devices,as well as a growing demand for marine resources and marine scientific research,the underwater target laser detection and imaging technology with high precision and high frame rate have become one an important mean of marine exploration.Streak tube imaging lidar utilizes the space-time resolution of a high-speed streak tube camera to enable high-resolution,three-dimensional imaging of the illuminated scene.In order to realize the underwater laser imaging with long distance and high detail resolution,based on previous study of the streak tube imaging lidar system,this paper initially analyzed the influence of the underwater environment on the imaging,then three-dimensional imaging experiment was conducted on the indoor target,finally designed and optimized the target three-dimensional reconstruction algorithm based on the streak images obtained in the experiment,the main job description is as follows:1.Based on the laboratory self-designed streak tube laser lidar system with a streak tube camera as its core device,the theoretical analysis is carried out mainly within three aspects,which includes the streak tube camera(its structure,imaging principle and key parameters),streak tube laser lidar system(components and the working principle of each part)and the threedimensional reconstruction algorithm.2.A streak tube laser imaging lidar system has been built with group members for underwater three-dimensional imaging in this paper.This system is based on underwater laser transmission characteristics and imaging requirements,and it is composed of devices suitable for underwater environment,which includes high repetition frequency pulsed laser with a wavelength within the seawater window,the trigger,and the high frame rate digital CCD camera etc.3.The effects of water surface wave and water volume on imaging have been simulated and analyzed.First,different kinds of water surface waves and the wave distortion were simulated,then combining modulation transfer function of water volume with water surface wave to simulate the image through water and air-water surface.4.A three-dimensional indoor target imaging system was built to conduct the threedimensional imaging experiment with group members,according to the features of the typical target(ladder target)streak images obtained from the experiment,the traditional feature extraction algorithm was improved,and a three-dimensional reconstruction algorithm based on Gaussian fitting was designed,which achieves a distance resolution of 30 cm in the depth direction and reduces the distance reconstruction error to less than 10%.On this basis,the convolution neural network was used to optimize the Gaussian fitting algorithm to ensure the reconstruction effect and improve the efficiency of the algorithm.The experiment results and reconstruction results verify the rationality and feasibility of the system scheme and threedimensional reconstruction algorithm,respectively,and provide a priori basis for the underwater target imaging in the future.