Research On 3D Reconstruction Of Laser Point Cloud Based On Incomplete Vehicle

With the continuous improvement of China’s car insurance,the requirements of the people’s market services have continued to improve.Automobile vehicle information inspection and registration as a method that effectively protects public safety,and can effectively prevent some of the safety hazards caused by some owners to make illegal modifications.The traditional manual testing method is low,and it is impossible to meet the needs of the market.The use of laser radar using laser radar using laser radar and three-dimensional reconstruction.It can provide both the vehicle point cloud model for the data platform,but also provides basic data for intelligent inspection vehicles,enriched the contents of vehicle detection.First,a vehicle feature extraction and reconstruction system based on laser radar is proposed.The system uses laser scanning radar to deactivate vehicle point cloud data within the inspection area;in point cloud data pretreatment,a adaptive combination is proposed.Filter algorithm,by adaptive change filter parameters,noise on the point cloud.Aiming at the lack of vehicle point cloud characteristics,a RANSAC-based hurt body feature repair algorithm is proposed,and the vehicle-symmetric mathematical characteristics are used to replenish the vehicle characteristics.In the point cloud ratio algorithm,by analyzing the shortcomings of the traditional algorithm,a three-segment point cloud registration mode based on an annealing algorithm is proposed.The first paragraph is a point-to-point registration,use the quick point characteristic histogram(FPFH)to look for feature point pairs,then use the analog annealing algorithm iterative feature point match pair;the second segment is characteristic registration,the first segment registration Input,construct feature point cloud,use KD-Tree search optimization ICP algorithm registration;third paragraph is the point cloud registration,and finally use ICP algorithm registration for overall clocks.After obtaining a relatively complete vehicle point cloud,the key points required for the intelligent check robot check process are classified.According to the position and characteristics of the vehicle characteristics of the key point,the characteristic segmentation algorithm based on the vehicle point cloud includes,including the interlimase-based door feature division algorithm,the edge fitting wheel feature segmentation algorithm,based on the region-based engine Gate Special Segmentation Algorithm.Effectively extracts the feature point clouds of different parts of the vehicle,which subsequently extracts the key points required for the verification process separately.After dividing the characteristics of the vehicle point cloud,threedimensional reconstruction is performed on the vehicle profile,and the Poiss Song Surface reconstruction algorithm and the greed triangulation algorithm are introduced.A three-dimensional reconstruction result surface hole repair scheme is proposed for the lack of experimental results of the two algorithm.Finally,the intelligent verification vehicle system has been proposed,and the venue and platform of the experiment is proposed,and the intelligent inspection car key software platform has been designed to complete the simulation experiment and experimental verification of vehicle characteristics extraction and reconstruction.The results prove the effectiveness of vehicle characteristics and reconstruction systems based on laser radar proposed in this paper.