激光雷达
噪音(视频)
中值滤波器
暗框减法
计算机科学
图像噪声
光子计数
滤波器(信号处理)
计算机视觉
人工智能
噪声测量
降噪
图像处理
光学
遥感
物理
光子
图像(数学)
地理
作者
Yongqiang Chen,Yan He,Fei Wang,Fanghua Liu,Chongmiao Jiao,Shouchuan Guo,R. K. N. D. Rajapakse,Weibiao Chen
出处
期刊:Twelfth International Conference on Information Optics and Photonics
日期:2021-11-01
摘要
Single-photon counting Lidar (SPL) systems using Geiger mode Avalanche Photo Diode (Gm-APD) arrays are more sensitive than traditional linear mode Lidar systems and are capable to detect the sparse target-returned photon less than single photon. However the single-photon sensitivity of SPL system also make it susceptible to solar background light which is a major noise source of SPL data. The relatively high noise level of SPL systems poses a significant challenges to the noise filter processing of measured data. In this paper, two image based noise filtering methods: K-Nearest- Neighbor (KNN) filtering method and Single Frame Histogram (SFH) filtering method were proposed, to reduce the noise points in Gm-APD array Lidar data. In these methods, noise points were removed through raw image data processing. We count the number of corresponding time of flight data points in single frame image and remove the noise points from signal through a predefined threshold. The noise filtering results of the two proposed methods were analyzed and compared based on raw data obtained from our 64×64 Gm-APD array Lidar imaging experiment. The noise filtered image results show that more than 90% of the noise points in single frame data has been removed. Finally, the noise filtered image data was further processed to get the cleaned 3D images. The results indicates that the proposed imagebase noise filtering methods is suitable for the noise reduction processing of our (Gm-APD) array Lidar data.
科研通智能强力驱动
Strongly Powered by AbleSci AI