Noise modeling, evaluation and reduction for the atmospheric lidar technique employing an image sensor

Atmospheric lidar signal, measured by the lidar technique employing an image sensor, suffers from sunlight shot noise, dark current noise, readout noise, and fixed pattern noise (FPN) of the image sensor. A noise model has been established to describe the noise characteristics and verified by evaluating lidar signals measured by an 808-nm Scheimpflug lidar system employing a CMOS image sensor as the detector. The sunlight shot noise and the photo-response non-uniformity (PRNU) noise that is one of the FPNs are found to be the primary noise sources of the lidar signal. The PRNU noise ratio is strongly dependent on the total illumination intensity of the image sensor and is minimized under high-light-level conditions. Thus, automatic exposure is suggested to achieve the best signal-to-noise ratio. Three different digital filters are employed to suppress the noise of the lidar signals, among which the Savitzky-Golay filter achieves the best performance. Moreover, a signal resampling method is proposed to improve the SNR for the near-range lidar signal. This work provides an in-depth understanding of the noise characteristics and proposes dedicated signal processing methods for atmospheric lidar techniques employing image sensors as detectors.