Time - Dependent Single Photon Counting Lidar Signal Analysis

LiDAR is an active 3D imaging system that works through a variety of work modes. Compared to the microwave radar, LiDAR has many merits, for example higher angular resolution, smaller divergence angle and other advantages, and has been widely used in various fields. Time-correlated single photon counting technique (TCSPC) is a discrete random signal sampling techniques, the features can be extracted by analyzing the statistical information, analysis features. TCSPC LiDAR based on the pulsed laser ranging, both reduces the laser intensity and increase the measure range and the ranging accuracy. TCSPC is a new discrete sampling techniques, and has the ability of photon counts, and a very high time resolution. With the recent relevant information, a reasonable reliable detection model is established; then the practical experiment was done to extract the accurate distance.By the detection model, the photon histogram has strong randomness, so a new approach is proposed for signal processing, based on the Markov Chain to extract the peak position, peak amplitude and the noise, then it is easy to get the accurate distance. In the data model, each parameter obeys the prior distribution which remains unknown, the preset posterior distribution and threshold techniques could effectively increase the computing speed. Metropolis algorithm can effectively reduce the amount of computation, as the result, the problem can be solved faster. For such combination problem, the parameters were updated by the Metropolis algorithm in cycle-mode, finally use the Bayesian inference for the optimal solution in the global approximation solutions.We propose a fast-approach STMCMC (Simulated Tempering Markov Chain Monte Carlo) for LIDAR signals with multiple returns, in order to obtain a complete characterization of a 3D surface by the laser range system. STMCMC is used to explore the spaces by the preset distributions instead of the prior distributions. Added active intervention tempering makes the Markov chain mix better through the temporary expansion of the solutions. The added step keeps the operation under control and yet retains the Markov characteristic of the operation. The theoretical analysis and the demonstrations on the practical data show flexible operation, and the parameters can be estimated to a high degree of accuracy.