Research And Implementation On Basic Algorithms Of 3D Reconstruction From Medical Images

3D reconstruction based on slice-images data is a main topic of visualization of 3D data sets. With the development of computer technology and computer graphics, medical imaging technology has improved greatly. 3D reconstruction based on medical slice-images has become a new step of the computer technology. 3D reconstruction from medical images is a multi-disciplinary subject. It relates to the subjects of digital image processing, computer graphics and some related knowledge of medical. So the study on basic algorithms of 3D reconstruction from medical images has important signification on science and worthiness in practical application.The purpose of this paper is to improve the basic algorithms of 3D reconstruction and to enhance the efficiency of the algorithms to meet the characteristic of medical images. Marching cubes (MC) algorithm is a classical algorithm of 3D reconstruction. An improved algorithm is presented in this paper for the disadvantage of the bad efficiency and the big quantity triangles of the traditional MC algorithm. In this method the isosurface can be protracted directly based on the function value distributing of the cube vertices. The process to look up the configuration style in the predefined table is avoided. Meanwhile a dissection method to solve the ambiguity problem is developed. An example of 3D reconstruction from medical images is presented and the results show that this algorithm accelerates the speed of traditional MC algorithm.Reconstruction of 3D objects from serial 2D contours will be applied widely in visualization of 3D data sets. An improved contour extraction algorithm on grid sequence is presented in this paper. It is named contour extraction algorithm based on region growing. The region growing method is used to speed up searching boundary grids and to avoid searching all of the grids. The experiment results show that the efficiency is greatly enhanced. The methods of reconstruction 3D objects from single branching 2D contours are presented. The problems of correspondence and branching in reconstruction 3D objects from multiple branching 2D contours is discussed particularly. A scheme for those problems is presented in detail. An example is made to show the process of reconstruction 3D objects from 2D contours and to prove thepracticability of the improved contour extraction algorithm.