Research On 3D Scene Reconstruction Algorithm Based On Single-line Laser

With the development of computer vision,three-dimensional scene reconstruction technology has been favored by many domestic and foreign researchers.In recent years,it has gradually become an important part of computer vision.The three-dimensional scene reconstruction technology provides reliable environmental information for mobile robots,and is the basis for mobile robots to achieve autonomous movement,plan paths,and avoid obstacles.The extensive application of three-dimensional reconstruction technology brings great convenience to people's daily life.This paper has designed a hand-held device that uses a single-line laser sensor and mounts it on a turntable that can rotate continuously to make a three-dimensional reconstruction of the scene.The three-dimensional scene reconstruction method implemented in this project can not only obtain accurate motion estimation,but also satisfy the real-time calculation.At the same time,this method is suitable for a variety of environments.The main idea is to decompose the complex SLAM problem into two algorithms.One algorithm computes laser odometer information at high frequency,and the other algorithm performs point cloud matching at low frequency.This topic focuses on the content of the following three parts:(1)Extract and screen feature point from single-line laser.Firstly,the principle of laser sensor measurement and the error in the measurement are analyzed.In the real scene,the data collection and pretreatment are the basis of follow-up work.The feature points were extracted and screened for the laser sensor measurement data,and the implementation process of the extraction feature algorithm was introduced.(2)Method for calculating laser odometer based on feature points.The method of describing the association between the feature points and the mathematical principle of the LM method for solving the least squares problem are described.Find the corresponding relationship between the feature points in two consecutive laser scan data,and then obtain the equation equations between the feature points according to the geometric relationship,and finally use the least squares method to obtain the information of the laser sensor odometer.(3)Optimization of laser pose based on mapping algorithm.After obtaining the odometer information of the laser sensor,the latest point cloud is converted into a world coordinate system according to the coordinate conversion formula,and a partial map point cloud is obtained,and is matched with the global map point cloud to obtain the local point cloud in the world coordinate system.The optimal position and posture,the current point cloud will be converted to the world coordinate system and join the map point cloud,update the map point cloud.