A Single-Photon Lidar Reconstruction Method Based on the Similarity Feature of the Whole Echo

Single-photon lidar usually relies on the superposition of multi-frame data for the signal extraction and three-dimensional reconstruction. However, how to reconstruct clear images with fewer frames and relatively low signal-to-noise ratio (SNR) is the main challenge of current research. The traditional peak method recomposes the image with the problem of blurring the edge of the target background area when SNR is relatively low, which limits the real-time efficiency of single-photon lidar imaging. Based on the similarity between the signal echo and the full pixel echo, this paper proposes the full echo similarity feature to distinguish the target and the background echoes. The proposed feature can achieve a better imaging effect without complex preprocessing by calculating the product and imaging the weight curve of each echo and the full echo. The experimental results show that under low SNR, compared with the characteristics of variance, peak value and peak location, the aliasing degree of full echo similarity on the two types of the target and the background echoes is lower. When counting 500 to 10,000 frames of data, the standard deviation between the characteristic values of the target and background area is less than 0.4. The proposed feature is less sensitive to the number of reconstructed frames. It can extract the stable edge pixels at 800 frames, which is much less than the number of frames required for variance, peak value and peak location feature.