一种基于动态点云的三维监控与压缩系统  

Three-Dimensional Monitoring and Compression System Based on Dynamic Point Cloud

作　　者：万杰[1] 廖燕俊 朱映韬 罗承明 陈建[1,2] WAN Jie;LIAO Yanjun;ZHU Yingtao;LUO Chengming;CHEN Jian(School of Advanced Manufacturing,Fuzhou University,Quanzhou 362200,China;School of Physics and Information Engineering,Fuzhou University,Fuzhou 350116,China)

机构地区：[1]福州大学先进制造学院,福建泉州362200 [2]福州大学物理与信息工程学院,福建福州350116

出　　处：《电视技术》2023年第12期34-39,共6页Video Engineering

基　　金：国家自然基金(62001117)。

摘　　要：动态点云能准确表达三维空间位置关系,相较于二维影像,在目标检测、人脸识别以及可视化等方面具有更好的表现,因此动态点云在视频监控领域具有较大应用前景。基于所提出的改进的动态点云编解码框架,实现一种基于动态点云的三维实时监控与压缩系统。首先,通过ZED 2i双目相机进行点云视频获取,以Jetson Nano作为数据处理器,应用基于统计学的滤波算法实现离群点与噪声的去除。其次,依据监控场景的静动特性进行前后景分割,分别应用提出的改进算法和PCL库的压缩算法进行编码。实验表明,在监控场景下,获取的动态点云序列取得了较好的主观效果的同时,实现了点云数据的高效压缩。Dynamic point cloud can accurately express the three-dimensional spatial position relationship,and has better performance in object detection,face recognition and visualization than two-dimensional image.Therefore,dynamic point cloud has a greater application prospect in the field of video surveillance.Based on the improved dynamic point cloud codec framework,a 3D real-time monitoring and compression system based on dynamic point cloud is implemented in this paper.Firstly,the ZED 2i binocular camera was used to obtain the point cloud video,Jetson Nano was used as the data processor,and a statistics-based filtering algorithm was applied to remove the outliers and noise.Secondly,according to the static and dynamic characteristics of the monitoring scene,the front and back scenes are segmented,and the improved algorithm and the compression algorithm of PCL library are respectively applied to encode.The experiment shows that the dynamic point cloud sequence obtained in this paper achieves good subjective effect and realizes efficient compression of point cloud data under the monitoring scenario.

关 键 词：动态点云压缩 点云分割 三维监控 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]