Research On Gm-APD Lidar Transmitting Algorithm Based On Model Parameter Estimation

The foggy environment is complex and it is difficult to identify the target information,so it is necessary to process the foggy image to recover the real scene information.Laser transmitted in fog is often absorbed and scattered by fog droplets,which weakens the target echo and makes it difficult to extract target information.To solve this problem,based on the Geiger mode trigger detection model and the backscattering of fog,a Gm-APD lidar parameterized model through fog imaging method was proposed.The model parameters were optimized,the original data were fitted,the backscattering of fog was removed,the target position was extracted,and the3 D image of the target was reconstructed.Firstly,the trigger probability model of Gm-APD lidar is introduced,the backscattering model of fog in closed indoor environment is studied,and the parameterization model of Gm-APD lidar fog penetration imaging is proposed.The single-parameter optimization fog penetration reconstruction algorithm for k is determined,and the double-parameter optimization fog penetration reconstruction algorithm for ? and k is determined simultaneously.Secondly,the problems existing in the fog-permeating imaging of the single estimation method were analyzed,and the fog-permeating strategy of the calibration factor was proposed.The trigger times of the data histogram were arranged in descending order.The average of the first n trigger times was selected as the normalized coefficient,the retention of the target position was counted,and the optimal value of n was determined to be 4.The original data were subtracted from the fitting model to remove the fog peak,and the matched filtering was carried out on the target peak signal.The filtering effect with different Gaussian widths was analyzed,and the peak value of the filtering signal was taken as the distance value of the target reconstruction.The least square single parameter optimization algorithm of fog permeation reconstruction is proposed,and the fitting effect is analyzed by using the sum of error squares and the determinate coefficient.Indoor Gm-APD lidar fog penetration imaging experiment was designed,and a platform was built to carry out experiments with different smoke concentrations and different target distances,to verify the reconstruction effect of the least square single parameter optimization fog penetration reconstruction algorithm,and the target recovery can reach 30.22%.Finally,put forward the double parameter optimization genetic algorithm and particle swarm double parameter optimization through fog reconstruction algorithm,the maximum likelihood estimate of k value as initial value,and search method is used to determine the k parameter space,the indoor experimental measured attenuation coefficient range as the parameter space,? a double parameter optimization space,analysis of fitness evolution curve of the convergence and stability of the algorithm.Compared with the single estimation method,the least squares single parameter optimization algorithm,genetic algorithm double parameter optimization algorithm and particle swarm optimization double parameter optimization algorithm respectively improved the restoration degree of the reconstructed target by 28.68%,45.43% and47.97%,and the signal-to-noise ratio increased by 0.4176,0.8702 and 0.9651.The relative mean ranging error is reduced by 0.172,0.2093 and 0.2294.In the outdoor environment with visibility of 2 km,the distance image restoration can reach 17.21% by using particle swarm optimization algorithm for the reconstruction of buildings at 1500 m.When the intensity image is processed at the image level,the structural similarity of the target is reduced by 0.0844,which indicates that the single-scale Retinex algorithm has limited image dehazing effect.