Theoretical And Experimental Research On Three-dimensional Imaging And Single Photon LiDAR

With the development of high sensitive single photon detection,high-precision timing and high speed data processing technology,single-photon lidar has developed rapidly in the field of active imaging.In this work,theoretical and experimental studies on time-correlated single photon counting lidar system are caried out.A single photon radar system with two modes of conventional acquisition and adaptive photon intensity acquisition has been built for depth estimation and depth algorithm reconstruction.1、Experimental study of time-correlated single photon counting lidar.A biaxial TCSPC lidar system has been built for depth estimation of indoor close-range targets and three-dimensional imaging experiment.The half-width of the instrument response function of the single-photon lidar system is 340 ps.A photon signal acquisition mode with adaptive photon intensity is developed by using the raster scanning method.This mode first prescans the whole to get the required photon acquisition pixels,then fine scanning to get the photon signal.This mode can effectively reduce the acquisition time when the reflectivity of the detected target is high.2、Depth estimation of rising edge based on instrument response function.Accurate depth estimation plays an important role in single-photon lidar.In the case of a small number of return photons,the estimation accuracy of the traditional algorithm is greatly reduced.Based on the traditional method of fitting depth estimation for instrument response function,this work proposed a depth estimation method based on the rising edge of IRF fitting function.This method has better distance accuracy and stability than traditional IRF curve fitting and maximum likelihood estimation method.3、Research on deep reconstruction of regularized denoising and deconvolution.Because pixel-by-pixel depth estimation is a depth estimation method for a single pixel point,the relationship between adjacent pixel points is not considered.Central pixel are affected by overlays of nearby pixel points when the spot size illuminated on the target surface is lager than the actual interval between pixles.In this work,single photon detection physical model has been built and spot size has been measured by charge coupled device.A regularization algorithm is used to solve the deconvolution problem and image denoising to get the depth map of the target.Compared with Reversible Jump Markov Chain Monte Carlo algorithm and Maximum Likelihood algorithm,proposed method has better smoothing effect and more detailed feature.