Denoising Method Based on EMD-CPSD for Low SNR Signals From Velocity Measuring Underwater Laser Light Screen

For the measurement of velocity using underwater laser light screens, the light screen passing through signal (LSPTS) is often drowned in noise due to the water scattering and water wave disturbances, and thus the signal-to-noise ratio (SNR) of LSPTS is usually lower than 5 dB, which leads to the failure of the cross correlation algorithm and the inability to extract the time interval between two screens, making it difficult to measure the underwater velocity. To improve the SNR of LSPTS, the empirical mode decomposition-cross power spectral density (EMD-CPSD)-based denoising method is proposed in this article. EMD is used to decompose the signal into a series of intrinsic mode functions (IMFs) and residual signal, calculating the frequency when the CPSD amplitude of the IMFs and the original signal reaches its peak, and then the IMF with the first minimal frequency is selected as the segmentation IMF. The denoised signal is reconstructed with the low-frequency IMFs and the residual signal after segmentation. The denoising experiments, underwater velocity experiments, and comprehensive evaluation results show that the method is more effective in improving SNR compared with EMD-correlation coefficient (CORR), EMD-Hausdorff distance (HD), and EMD-consecutive mean square error (CMSE) methods, which verifies the effectiveness and practicality of the method.