A novel lidar signal-denoising method based on IPSO-VMD-ISVD algorithm

Abstract

As atmospheric lidar signals are susceptible to various noises, the quality and accuracy of the signals are seriously affected. For this problem, this paper proposed a joint denoising method combining variational mode decomposition (VMD), improved particle swarm optimization (IPSO), and improved singular value decomposition (ISVD), named IPSO-VMD-ISVD. The parameter-adaptive IPSO algorithm was employed to optimize the key parameters of VMD (decomposition level K, penalty factor α), effectively addressing the challenge of determining the optimal parameter combination in the VMD algorithm. Based on the correlation coefficient threshold method, intrinsic mode functions were selectively filtered. Subsequently, a segmented denoising algorithm, leveraging the cyclic matrix approach of ISVD, was applied to further suppress low-frequency and high-frequency noise in the signal, yielding the final denoised result. Experimental results indicate that, for both simulated signals and actual lidar signals, the IPSO-VMD-ISVD method demonstrates superior performance compared to empirical mode decomposition, sparrow search algorithm-based VMD modify to wavelet transform (VMD), and particle swarm optimization-based VMD. This superiority is evidenced by improved evaluation metrics, including higher signal-to-noise ratio and correlation coefficient, as well as lower root mean square error and percent root mean square distortion. This method provides an effective and practical denoising solution for lidar signals.