Reconstruction method of 128 × 256 array single photon Lidar based on multi-domain stability feature fusion

Abstract

Under low-light conditions, random light distribution and non-uniform pixel sensitivity reduce both the correlation and differences among pixels, while unstable intensity information significantly impairs the detection capability of Geiger-mode avalanche photodiode (GM-APD) arrays. To address these challenges, a method based on multi-domain stability feature fusion is proposed. This approach utilizes a distance layer decomposition model to break down the global problem into localized sub-problems, effectively suppressing background noise through the fusion of stable features. Additionally, the Multi-scale Algorithm (MSA) was enhanced to selectively recover missing pixels and improve target reconstruction while preserving details. In imaging experiments conducted on targets under low-light conditions at night within remote, complex scenes, when the photon number was 0.0068 per pixel, the proposed method improved the Peak Signal-to-Noise Ratio (PSNR) of the reconstructed images by more than 12 dB compared with the Non-local MSA. It significantly promotes the development of GM-APD lidar for all-time applications.