Adaptive digital filter for the processing of atmospheric lidar signals measured by imaging lidar techniques

The lidar signal measured by the atmospheric imaging lidar technique is subject to sunlight background noise, dark current noise, and fixed pattern noise (FPN) of the image sensor, etc., which presents quite different characteristics compared to the lidar signal measured by the pulsed lidar technique based on the time-of-flight principle. Enhancing the signal-to-noise ratio (SNR) of the measured lidar signal is of great importance for improving the performance of imaging lidar techniques. By carefully investigating the signal and noise characteristics of the lidar signal measured by a Scheimpflug lidar (SLidar) based on the Scheimpflug imaging principle, we have demonstrated an adaptive digital filter based on the Savitzky–Golay (S–G) filter and the Fourier analysis. The window length of the polynomial of the S–G filter is automatically optimized by iteratively examining the Fourier domain frequency characteristics of the residual signal between the filtered lidar signal and the raw lidar signal. The performance of the adaptive digital filter has been carefully investigated for lidar signals measured by a SLidar system under various atmospheric conditions. It has been found that the optimal window length for near horizontal measurements is concentrated in the region of 90–150, while it varies mainly in the region of 40–100 for slant measurements due to the frequent presence of the peak echoes from clouds, aerosol layers, etc. The promising result has demonstrated great potential for utilizing the proposed adaptive digital filter for the lidar signal processing of imaging lidar techniques in the future.