The lidar denoising algorithm based on an improved correlation parameter of ensemble empirical mode decomposition

Under the condition of weak signal of photon-counting lidar and strong noise of solar background, the signal is completely submerged by noise, resulting in the detection of multiple peaks through photon-counting entropy. Consequently, the distinction between signal and noise may become difficult, causing the significant fluctuation in ranging error. To address this issue, we propose the lidar denoising algorithm based on an improved correlation parameter of ensemble empirical mode decomposition, including the coarse denoising stage and recognition stage. In the coarse denoising stage, the method of ensemble empirical mode decomposition is primarily used for extracting and eliminating the noise components from the signal. To identify noise components, we propose an improved correlation parameter based on the combination of first-order linearity and second-order nonlinearity fitting using the least squares algorithm. In the recognition stage, the photon-counting entropy is further utilized for anti-noise and identifying the target signal. According to the simulation and experimental analysis, the ranging error of our proposed method are less than 5 and 30 cm, respectively. When compared with the denoising algorithm of photon-counting entropy, the average ranging accuracy is enhanced by 74.69% and 74.42%, respectively. Meanwhile, in comparison to other algorithms, it also possesses superior capabilities.