基于改进欧式聚类镜像反射噪声的去除  

Removal of Mirror Reflection Noise Based on Improved Euclidean Clustering

作　　者：杨勇[1] 刘德儿[1] Yang Yong;Liu De'er(School of Civil and Surveying Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,Jiangxi,China)

机构地区：[1]江西理工大学土木与测绘工程学院,江西赣州341000

出　　处：《应用激光》2025年第2期179-187,共9页Applied Laser

摘　　要：激光雷达获取点云数据时,受到玻璃、镜子等反射物体的影响,会产生镜面反射,从而产生镜像反射噪声。由于镜像反射噪声与对应实际点云存在较多相似性特征,传统去噪方法难去除该类噪声点。对此,基于两者的相似性与差异性特征,提出了镜像反射噪声的识别和去除方法。首先将场景点云数据分割成若干点云块,通过随机采样一致性算法提取出镜像对称面,接着判断点云块到镜像对称面的欧式距离初步筛选出镜像反射噪声,通过三维点云配准和二维深度图像的相似性检测准确地识别出镜像反射噪声,最后利用点云凸包计算密度值结合欧式聚类去除镜像反射噪声。实验结果表明,所提方法能准确识别出镜像反射噪声并将其有效地去除。LiDAR point cloud data acquisition is often affected by reflective objects such as glass and mirrors,which produce specular reflections and introduce mirror reflection noise.Due to the similarities between mirror reflection noise and the corresponding physical point clouds,traditional denoising methods struggle to effectively remove such noise points.To address this challenge,this paper proposes a method for identifying and removing mirror reflection noise based on the similarity and difference between the two.First,the scene point cloud data is divided into several point cloud blocks,and the mirror symmetry plane is extracted through the random sampling consensus algorithm,and then the Euclidean distance from the point cloud block to the mirror symmetry plane is judged to preliminarily screen out the mirror reflection noise,and the 3D point cloud registration is performed The similarity detection with the two-dimensional depth image accurately identifies the mirror reflection noise,and finally the point cloud convex hull is used to calculate the density value combined with Euclidean clustering to remove mirror reflection noise.The experimental results show that this paper can accurately identify the mirror reflection noise and remove it effectively.

关 键 词：点云去噪 镜像反射噪声 点云配准 欧氏聚类 密度分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]