Research On 3D Geometry Modeling Technology Based On CT Image

With the progress and development of modern science and technology,interdisciplinary has become the trend of development.In medical field,medical image data is two-dimensional slice sequence image.But with the introduction of computer technology,three-dimensional(3D)geometric modeling of medical image has become the trend.With the application of 3D modeling,simulation surgery,personalized medical device design and manufacture,etc.have become possible.It is having more and more application in many fields.This paper studies the basic technology and application of 3D geometric modeling.Using skull CT images as the data,three-dimensional reconstruction of skull is realized by marching cubes method.Aiming at the low efficiency of marching cubes method in the case of large CT data,we accelerate the reconstruction process by removing the empty cubes through the region growing and avoiding repeated interpolation.Compared with the original MC algorithm,it successfully speeds up 40%-60%.Both the field of computer graphics and practical engineering applications require that the 3D model should be manifold.This paper first improves non-manifold data structure AHF and uses the list to realize the non-manifold data structure LHF,which supports dynamic additions and deletions.Then,the paper proposes a method to repair the non-manifold model by inserting vertices in the middle of the edge.This method can properly repair the non-manifold model and keep the edge connectivity of the model with only a small amount of model complexity increment.In the end,the paper studies the technology of section contour extraction.According to the characteristics of the plane intersection,a spherical bounding box based on R tree is proposed to index the 3D model.Experiment shows that in plan intersection,our method is more efficient than R*tree.For the mega scale 3D model,the section contour can be successfully extracted in 80ms.In this paper,we study several commercial level issues related to the 3D geometric modeling of medical images.In geometric modeling,we improve the efficiency of MC algorithm so that the voxel modification can get faster response;in non-manifold repair,we realize a non-manifold data structure supporting dynamic modification and propose a non-manifold repair method which can keep the edge connectivity,making the 3D printing and other applications possible;in section contour extraction of 3D models,we improve the index method and the medical staff can get the section counter of the model in real time to achieve precise positioning and other operations.