LiDAR Waveform Decomposition Algorithm And Its Application In 3D Imaging

LiDAR three-dimensional imaging is a common technique for three-dimensional reconstruction of real scenes using a LiDAR system combined with scanning or array devices.It has the characteristics of high spatial resolution and can realize the imaging of targets under cluster leaves.Against this background,this paper proposes an echo signal decomposition method based on exponential function fitting system waveforms for the above-mentioned applications of LiDAR(Exponential Decomposition,ED),in order to effectively improve the information extraction capability and positioning accuracy of occluded targets.Based on data investigation,the paper outlines the basic principles of LiDAR ranging and three-dimensional imaging,sorts out the research status of LiDAR waveform processing,and introduces a deconvolution method to provide accurate echo number for laser echo waveform decomposition.It is estimated that the accuracy of several commonly used waveform deconvolution methods to obtain the number of echoes is compared and analyzed through simulation calculations.It is found that the L1 regular method waveform deconvolution is most suitable for determining the number of waveforms in laser echo decomposition.A method of laser echo signal decomposition based on exponential function fitting system waveform was proposed,and combined with L1 regular method waveform deconvolution to determine the number of waveforms,a waveform decomposition program was written in MATLAB software environment.In the experiments,a digital radar full-waveform acquisition technology was used to establish a LiDAR scanning imaging experiment system,and a multi-target ranging experiment and a three-dimensional imaging experiment of targets under cluster leaves were performed.Using the method proposed in this paper to process the laser echo waveform to determine the multi-target distance.Compared with the traditional Gaussian decomposition algorithm,lognormal algorithm and generalized normal algorithm,the ranging accuracy can be improved by 15%,50% and 36%.On this basis,the laser waveform signals of the three-dimensional imaging of the target under the cluster leaf are also decomposed,and the three-dimensional contour image of the target under the cluster leaf is extracted using Mean Shift method,and compared with the processing results of other three algorithms.The results show that this paper The proposed method can well recover the target box covered by leaves,has less stray points and stronger applicability.