Lidar Echo Signal Enhancement And Waveform Decomposition

Compared with the other imaging techniques (such as synthetic aperture radar, infrared, millimeter wave and visible light), lidar technology can be used to directly obtain high precision 3D data and to get more abundant image information (including 3D range image and gray image, etc.) and ability to anti disturbance (such as electromagnetic, background, and the sun). Because of this,3D lidar is becoming one of the important development of active imaging detection, receives more and more attention of our army and foreign armies. The 3D lidar imaging technology is limited by the size of the detector array and the laser and the scanning structure, it is difficult to obtain a further breakthrough in the scanning speed and precision.This paper introduces the lidar’s development in China and other countries, and introduces three-dimensional lidar’s ranging principle. Through the simulation and modeling of the system, the corresponding system model is established, it analysis the key characteristics of the system, the signal of noise environment is analyzed in detail. This is useful in signal processing algorithm presented in this paper and laser radar analysis and modeling, mainly including laser emission system, atmospheric transmission, objects and receiving and detecting system.This paper presents an improved de-noising algorithm of laser signal based on empirical mode decomposition, in the traditional denoising method, selecting different wavelet bases have different effects on the denoising results of wavelet denoising, it is necessary to choose the correct wavelet, so the noise removal is not adaptive; empirical mode decomposition the algorithm decomposes the signal into different intrinsic mode components, that the noise in high frequency intrinsic mode components, removing noise intrinsic mode components which will cause the loss of effective signal; FFT transforms the signal from the time domain into the frequency domain, thus it will result in the loss of its time-domain characteristics, there are some limitations for non-stationary signal that may lead to loss of signal. This paper presents an improved EMD algorithm based on processing the high frequency IMF, which no longer simply discard the noise component but to deal with it, so it reduces the loss of effective signal.Some random noise generated by random point and the burr needs to be smoothed, otherwise they will have bad effects on waveform decomposition. This paper introduces the Vondrak smoothing algorithm which is firstly used in astronomical data processing. The algorithm is suitable for nonlinear data, and the laser echo signal is a typical nonlinear data. In this paper, the principle of the algorithm and its derivation formula, and the selection of smoothing factor and the evaluation of the smoothing effect are introduced. The validity of the algorithm is verified by the analysis of data and compared with other algorithms, such as the five point three smoothing.The model in traditional LM algorithm of laser waveform decomposition algorithm is symmetric Gaussian model, however, the actual waveform is not and it is not symmetric. So, traditional algorithm which gets the pulse width and other parameter by obtaining the symmetric inflection point is not accurate, at the same time, LM’s accuracy is related to the initial value. In this paper, a decomposition of lidar echo waveform based on the particle swarm optimization is proposed, in which, we set a threshold according to the background noise and get the number of single waveform by peak detection, then we obtain the pulse width and delay time by PSO algorithm, lately by using LM algorithm, the parameter’s fitness is improved.Through three kinds of target model, Gauss model, slope model and building model’s simulation and reconstruction, it proves that the algorithm is effective.