Research And Implementation Of3d Lidar Interactive Display And Pointcloud Analysis System

With the development of LiDAR technology,more and more enterprises and schools over the world have carried out research on 3D LiDAR.The principle of 3D LiDAR technology is to measure the angle and distance of objects around the LiDAR by reflecting laser,and convert them into point coordinates in space.Multiple points can form point cloud,and the research on pointcloud is helpful for environment perception.LiDAR technology has been applied to all levels of society: in military,LiDAR has been widely used in detection,guidance,measurement and other applications;in civil,LiDAR has been widely used in unmanned driving,mobile robots,auxiliary detection,scene building,virtual reality and other scenes,and the demand is expanding.In order to break the monopoly of foreign countries,we cooperate with domestic LiDAR manufacturers to develop a 3D LiDAR pointcloud interactive display and pointcloud analysis system with independent intellectual property rights,which provides supporting algorithms for the demonstration,evaluation and later application expansion of LiDAR equipment independently developed by domestic LiDAR manufacturers,and promotes the development of 3D LiDAR research and application in China,It has important value and significance.This paper analyzes the needs of domestic 3D LiDAR cooperative manufacturers.The company's needs include real-time display of LiDAR,evaluation of LiDAR measurement accuracy and the final expansion of LiDAR applications.Therefore,a 3D LiDAR interactive display and pointcloud analysis system is developed,which mainly consists of five modules: basic module,display module,filtering and denoising module,accuracy evaluation module and object detection module.Before the three modules mentioned above,the basic part and special filtering operation part are added.The basic module is responsible for pointcloud transmission directly with LiDAR,and the file storage and LiDAR communication are studied in detail.The display module is responsible for the work of pointcloud visualization,so that pointcloud can present different colors according to different attributes,and develop different color coding methods,which can have an intuitive understanding of pointcloud attributes.The main purpose of the filtering module is to reduce the impact of noise on the subsequent pointcloud processing.According to different application scenarios,different filtering methods are developed: for the static scene of real-time display,a filtering method based on the point cloud space-time is proposed,which is successfully applied to the LiDAR noise reduction;for the accuracy evaluation,the random noise of pointcloud is eliminated through multiple measurements,which makes the evaluation The result is more accurate.The accuracy evaluation module mainly analyzes the radar error and proposes a standardized quantitative evaluation method of LiDAR pointcloud quality.By measuring the high-precision standard room data and comparing with LiDAR pointcloud,LiDAR standard deviation,system error,packet loss rate and other attributes are analyzed,which provides an important basis for the development and quality inspection of LiDAR.The object detection module integrates the deep learning network into the system.By processing the pointcloud data in the road,the vehicle in the road data is detected with three-dimensional object,which completes the object detection function of the pointcloud and expands the application scene of the system.Finally,by analyzing the processing flow of pointcloud,the organizational structure of each module is creatively designed.Using Open GL visual library function,based on the object-oriented C++ programming language,we call the python deep learning network module,and carry out the research and development of the module function under the MFC development framework,and write the pointcloud visualization and analysis system with interface.Through the systematic test of each module,the accuracy of the experimental results is verified,and the system is smooth and stable on the premise of perfect function.