Research On 3D Reconstruction Technology Of Vehicle Contour

With the rapid development of China’s automobile industry,people’s demand for automobile industry and related technologies is also increasing.At the same time,many car owners in order to seek greater interests,usually refit the vehicle to increase the carrying capacity,seriously threatening public security,so it is very important to regularly test the vehicle.In traditional detection,not only the efficiency of manual detection is too low to meet the increasing needs of the people,but also the content of automatic detection is too single and the detection ability is limited,only the length,width and height data of vehicles can be detected.Combining the above two points,the 3D reconstruction of vehicle is carried out.It can not only provide the vehicle point cloud model to provide basic data for the new technology,but also collect all parts of the vehicle data,greatly enriching the detection content.In this paper,a 3D reconstruction system of vehicle contour based on laser ranging scanner is proposed.The system uses laser ranging scanner to obtain the point cloud data passing through the vehicle.In the preprocessing,the improved grid method is proposed to establish the point cloud topology,and the point cloud data is simplified by combining the chord height difference resampling.The statistical method and bilateral filtering algorithm are used to process the point cloud data Row noise reduction,using image processing method to segment all parts of the vehicle.In the point cloud data restoration,an algorithm based on angular differentiation is proposed to remove the noise points of the window,calculate and correct the tire data by the way of mathematical geometry,repair the point cloud data depression at the lamp by the way of fitting,and complete the reconstruction of the window and the hole by using the improved eight neighborhood method combined with the radial basis method.In the surface reconstruction,the triangle generation algorithm based on region growth and the Poisson surface reconstruction algorithm based on fitting are compared.It is determined that the algorithm based on region growth is better when fitting the vehicle point cloud data of the unclosed surface.Finally,the algorithm effect of this paper is analyzed qualitatively and quantitatively by3 D reconstruction model and vehicle detection error.The experiment shows that the systemcan complete the three-dimensional reconstruction and dimension data detection of the vehicle under the condition that the vehicle is moving at a low speed of less than 5km / h,and the detection error is in line with the motor vehicle safety technology inspection items and methods(gb21861-2014)The measurement error of the motor vehicle outline size detection system shall not exceed the requirements of ±1% or ±20mm.It is very important to provide3 D reconstruction model of vehicle and overload detection.