Research On Three - Dimensional Imaging Method Of Fast High Resolution Lidar

Recently, one of the most commonly used methods to acquire scene depth is the light detection and ranging (LIDAR) technique. To get a 3-D image, existing raster scanning 3-D imaging systems scan the scene by moving a laser source in a transverse area and illuminate each point of the scene, get the returned reflections at the photodetector and using those data to estimate the depth of each illuminated point. Then, they put these estimated depths together to form a 3-D image. Obviously, depth estimation is a core component for 3-D imaging systems, especially at low power levels. In this thesis, we consider how to make the time of flight depth acquisition as accurate as possible at low photon counts, and obtain a clear 3-D image.In order to exactly estimate the scene depth, we firstly consider modeling the photon counting process for the photodetector and its received reflections with taking account of the effect of scene depth variation, where we formulated the received light intensity as the convolution of the source light intensity and another signal which we call the scene impulse response. And then, in order to tackle the problem that traditional LIDAR can’t process signals with multiple returns, we emphatically studied on two kinds of mainstream full waveform analysis algorithm based on the Markov chain-Reversible Jump Markov Chain Monte Carlo (RJMCMC) and Simulated Tempering Markov Chain Monte Carlo (STMCMC), and the latter is the improved version of the former at the operation time performance.In the end, for traditional 3-D imaging model based on the maximum-likelihood estimation algorithm, it’s impossible to directly and accurately determine the pixel wise signal-acquisition time or the scanning step length in the case that the scene properties are unknown. In order to tackle these problem, in this thesis we propose a photon-counting adaptive depth imaging method for direct-detection 3-D imaging LIDAR system. The latter theoretical analysis and experiments will demonstrate that our adaptive imaging method can more quickly and accurately estimate the scene depth and get a clear 3-D image, even in the presence of high background noise.