Enhancement of Weak Lidar Signal Based on Variable Frequency Resolution EMD

Abstract

A new high-resolution approach is presented for noise reduction in lidar signals using variable frequency resolution empirical mode decomposition. Avalanche photo diode (APD) output single-channel low signal-noise ratio (SNR) real data are decomposed into multi-scale components and reconstructed with less noise. The key is to have a fine scale signal in each component. Based on the non-symmetric signal and the symmetric noise on the timeline in APD data, the proposed approach is highly effective in distinguishing between signal and noise by choosing an appropriate frequency resolution, compared with the conventional multiscale method. Real examples are examined with real data, signal separation is obtained in the components, and the -22.4 dB low SNR has been enhanced by the component product.