Research On Noise Reduction Technology Of GM-APD Array Lidar Echo Signal

Geiger Mode Avalanche Photon Diode(GM-APD)array lidar is a new type of nonscanning photon counting radar,which can respond to the echo signal of single photon energy.It has extremely high detection sensitivity,making it possible to detect weak signals.It has been widely used in cruise guidance,topographic mapping,underwater detection and other fields.However,because GM-APD is a probabilistic device and the laser has the characteristics of rapid attenuation in the medium,the detected echo signal is too sparse,and a large amount of accumulation is needed to recover the target echo signal,which limits the application of lidar in fast imaging occasions.On the other hand,due to the high detection sensitivity of GM-APD,it is susceptible to noise interference,resulting in deviation of detection distance,which limits the application of lidar in precise positioning occasions.Therefore,this paper focuses on the problem of poor real-time performance and low accuracy of GM-APD array lidar imaging,and studies the noise reduction algorithm of GM-APD array lidar echo signal.Based on the detection principle of GM-APD array lidar system,this paper analyzes the noise characteristics of lidar.Through the photon distribution of lidar echo signal,the trigger probability distribution model of echo signal of GM-APD array lidar is established.Combined with probability density function,cumulative probability function and Mente Carlo random sampling method,the simulation of echo signal data is realized,which lays a foundation for further research on noise reduction algorithm.Aiming at the problem of poor real-time performance of GM-APD array lidar imaging,this paper studies the classical Kalman filter algorithm and the simplified Sage-Husa adaptive algorithm,and improves the time-related adaptive Kalman filter algorithm.By increasing the rationality test,the amplitude of the photon data set is judged,and the problem of abnormal filtering caused by outliers is solved.The simulation results show that the improved timecorrelated adaptive Kalman filter algorithm reduces the root mean square error by 62.10 %and the running time of single frame program by 46.07 % compared with the simplified SageHusa adaptive algorithm,which can achieve fast noise reduction.Aiming at the problem of low imaging accuracy of GM-APD array lidar,this paper performs noise reduction processing on echo signals based on Empirical Mode Decomposition(EMD).Based on the EMD time scale noise reduction method and the EMD adaptive soft threshold noise reduction method,the EMD modified threshold method is proposed.By modifying the threshold of each modal function instead of the correlation mode discrimination method,the problem of inaccurate filtering caused by the unstable selection of the boundary points of noise and signal is solved.The simulation results show that compared with the EMDmoving average algorithm,the root mean square error of the EMD correction threshold method is reduced by 11.80 % when the number of fixed frames is 200,and the peak signalto-noise ratio is increased by 12.86 %,which realizes the accurate acquisition of the target distance information.Finally,a 64 × 64 GM-APD array lidar imaging experimental platform was built,and laser active imaging experiments based on GM-APD array were carried out.The experimental results show that the improved time-correlated adaptive Kalman filter algorithm has a clear target contour after denoising,and the single frame running time is reduced by 66.93 % and83.7 % respectively compared with the classical Kalman filter algorithm and the simplified Sage-Husa adaptive algorithm.The EMD modified threshold method proposed in this paper has better target integrity after denoising.The peak signal-to-noise ratio is 39.8 %,39.2 % and28.3 % higher than that of EMD time scale denoising method,EMD-SG filtering algorithm and EMD-moving average algorithm respectively.Therefore,the proposed algorithm can effectively process the actual measured echo signal and improve the real-time and accuracy of GM-APD array lidar imaging.