Font Size: a A A

Fast Parallel Computing Of CT Simulation Projection Of Cultural Relics Based On UG Model

Posted on:2021-03-30Degree:MasterType:Thesis
Country:ChinaCandidate:Y X XuanFull Text:PDF
GTID:2428330611481925Subject:Software engineering
Abstract/Summary:PDF Full Text Request
CT(Computed Tomography)technology,that is,computer tomography technology.This technology can obtain the internal structure information of the object through tomographic image reconstruction without loss,and it has been applied to the digital modeling of cultural relics.The premise of the research on digital modeling of cultural relics based on CT is to obtain projection data,but the cost of industrial CT scanning is high,and it may cause damage to cultural relics in the process of acquiring cultural relics CT data.Using the UG(Unigraphics NX)model to intersect with the ray can provide ideal projection data for the study of cultural relics image reconstruction.Its disadvantage is the large amount of calculation,especially for high-resolution projection calculation.In response to this problem,this paper intends to carry out rapid simulation projection calculation research based on the UG model.The main research contents of the paper include(1)Aiming at the problem of slow simulation calculation speed of UG model,this paper proposes to voxelize UG model and use voxel model for simulation calculation.For the single-material cultural relic UG model,first read the size,density and other information of the model,and then set an outer bounding box that surrounds the UG model,and mesh the bounding box.Each grid represents a voxel unit.Divide the bounding box into individual voxels.For the number of voxels along the x,y,and z directions of the bounding box,use the UG API function to calculate the intersection of each layer,and then obtain the voxelization of each layer in turn.Data,and finally generate voxelized data for the entire UG model.For the multi-material cultural relic UG model,first read the material information of each part,and find the density value corresponding to the material in the database,then voxelize each part in turn,and accumulate the voxelization results of each part.Get the voxelized data of the whole model.Experimental results show that this method can achieve accurate voxelization(2)Based on the UG model voxelized data,cone beam CT projection simulation is realized First adjust the model to the simulation system,then voxelize the model,and use the trilinear interpolation method to calculate the projection data of the model.In each ray,first calculate the starting intersection point of the ray and the model,and start sampling at equal intervals from the starting intersection point.The density of each sampling point is calculated by trilinear interpolation to obtain the density value of each sampling point.After that,it is accumulated and multiplied by a fixed step to obtain the projection value of the ray.It is proved by experiments that when calculating projection images with a resolution of 512 512 pixels under 360 angles,the projection data generated by the voxel model has achieved an acceleration ratio of about 3 times compared with the traditional analytical method(3)In order to further improve the speed of simulation calculation,a fast calculation method of CT simulation projection of voxel model based on GPU acceleration is proposed.First,bind the voxel model data to the GPU's texture memory,and then map each thread index of the GPU to each detector unit to achieve parallel calculation of all rays.Within each ray,the projection data is calculated using the trilinear interpolation method,and the density value of each sampling point is calculated by the three-dimensional texture hardware interpolation function.Through experimental analysis,when calculating projection images with a resolution of 512 to 512 pixels under 360 angles,this method achieves an acceleration ratio of about 83 times compared to CPU calculation of trilinear interpolation on the premise of ensuring accurate projection data.
Keywords/Search Tags:digitization of cultural relics, UG model, voxelization, GPU acceleration, CT projection simulation
PDF Full Text Request
Related items