Improved Algorithm Of 3D Reconstruction Based On Point Cloud Data And Its Application In Coil Loading And Unloading

The wisdom and digital transformation of the steel industry is an inevitable trend for the continuous development of traditional industries.Intelligent steel mill logistics,as an important part of the smart factory,helps to improve the transportation efficiency and safety of steel coils and other products.Nowadays,the entry and exit of steel coils in steel logistics often requires manual guidance of unmanned cranes for steel coil loading and unloading,which is low in work efficiency and difficult to guarantee accuracy.The use of computers to identify coordinate information in the process of steel coil loading and unloading by improving detection means is one of the important basic tasks of the intelligent steel coil logistics.In this paper,the point cloud data are collected by 3D laser scanner on the surface of the target at a certain frequency and resolution.The position information of the target is accurately restored by the point cloud,which solves the problem of inaccurate detection of steel coil and vehicle in the process of coil logistics.In this paper,a three-dimensional laser scanner is used to collect point cloud data on the surface of the measured target at a certain frequency and resolution,and the position information of the measured target is accurately restored through the point cloud to solve the detection of steel coils and vehicles in the steel coil logistics process.For the collected 3D point cloud data,the preprocessing through filtering algorithm is used to filter the invalid background data,the radius filter to remove noise points,and the Voxel Grid filter to simplify the point cloud data.Then the boundary recognition algorithm is used to determine the effective point cloud boundary information,the random sampling consensus algorithm(RANSAC)is used to fit the parameters of the model,and the cylindrical coordinates of the steel coil are fitted according to the spatial geometric model by projection.When the steel coil is shipped out of the warehouse,according to the saddle plane of the carrier,the column coordinates of the steel coil placed on the saddle are calculated through the geometric relationship between the steel column and the plane;when the steel coil is in the warehouse operation,According to the surface parameter equation of the steel coil,the cylindrical coordinates of the steel coil can be calculated.In view of the problem that the processing effect is greatly affected by the quality of the point cloud,by improving the Voxel Grid filter,the point cloud can be simplified while retaining the feature details,and the moving least squares smoothing process can solve the problem of uneven surface.After experimental testing,the reconstruction method designed in the thesis can reduce the problem of reduced recognition accuracy due to laser scanner errors and environmental influences,and improve the robustness of coordinate position recognition,which has important reference significance for the precise positioning of products in steel logistics.