Three Dimensional Visualization Of Nuclear Environment Based On Point Cloud Modeling Method

With further shortage of energy,the proportion of nuclear energy as clean energy is increasingly growing.Nevertheless,it is with great danger applying nuclear power.So the nuclear industry robots are in great need.This study of a three-dimensional nuclear work environment modeling is based on the National Natural Science Foundation of China," Research on Adaptive Mechanism and Efficient Operation Method of Robot Oriented Nuclear Power Plant RCV "(61473113).The 3D map converts the real three-dimensional space into digital information to facilitate the identification of the robot.Failing to create accurate three-dimensional map is the primary problem that nuclear power robots may face on autonomous operation.Therefore,this paper analyzes the reconstruction process of 3D model from the establishment of camera platform and obtaining of the point cloud data,and improves its noise processing and point cloud stitching method.The main contents and innovations of this paper are as follows:Firstly,this paper describes the various camera models,and compares the binocular camera and Kinect depth camera,then selects final selection of Kinect camera as a point cloud access tool with the use of Open NI to drive Kinect camera in the VS2015 platform and Open CV visual library for camera calibration processing.and nuclear power simulation scene for three-dimensional data acquisition.Ultimately the scene of the three-dimensional point cloud data is generated.Secondly,the noise of point cloud image is analyzed.In order to improve the quality of reconstruction,point cloud data is filtered.Due to the continuity of the midpoint,the large edge noise,and the disturbed noise inside the cloud in the actual situation,the traditional image filtering algorithm cannot complete the denoising.In this paper,we use the best connected domain method to judge and remove the outlier,to calculate the normal vector of point cloud data and correct it.The improved edge filtering is used to accurately denoise the data.Thirdly,the preparation of the point cloud data from different view is carried out.The position transformation relation between different coordinate systems is analyzed.The traditional ICP algorithm and NDT algorithm are introduced,and the basic principle of each step is analyzed as well.On this basis,the optimal splicing strategy of point cloud stitching is put forward,and the method of extracting eigenvector points of point cloud data is adopted to improve the quality of point cloud rough stitching.And the constraint condition and the equilibrium scale factor are introduced in the process of precise splicing,which improves the effect of 3D point cloud reconstruction.Finally,the algorithm proposed in this paper is experimented in the simulation of nuclear power plant.The results show that the improved filtering algorithm and splicing algorithm used in this paper improve the quality of point cloud data processing and improve the precision of point cloud mosaic reconstruction,which can be effectively applied in the reconstruction of real scene.