Methods And Applications Of 3D Image Reconstruction

With the rapid development of science and technology, people demand more and more information of 3D world. 3D Visualization has been one of research focus of computer technology, it involves computer graphics, computer aided design, Human-Computer Interaction Techniques and other theories and techniques. 3D reconstruction has been widely applied in social life, such as medical imaging, geology and exploration, digital city and military applications. But with the generation of large number of image information, it is requested much higher 3D reconstruction technical level. So the research of 3D reconstruction that makes the efficiency and accuracy better and enhances the visual effect has practical significance.This paper focused on the problem of image processing, theory and technology of 3D reconstruction. We proposed a segmentation method for CT images of cerebral hemorrhage and a more accurate method of calculating the volume of cerebral hemorrhage. And based on the marching cubes algorithm, we proposed a fast iso-surface extraction method. The main work is as follows.Firstly, we proposed a segmentation method for CT images of cerebral hemorrhage. According to the characteristics of cerebral hemorrhage CT images, we recognized the characteristics of the image with edge detection and binarization, and removed part of noise in the image. With two region growing method, we segmented complete skull and blood from CT images. An experiment case demonstrated the proposed method has good effect.Secondly, according to the characteristics of computed tomography scans, we proposed a cascading style's three-dimensions modeling, it used column method and truncated pyramid method calculate the blood volume. Compared with the coniglobus formula, this methods calculated the amount of bleeding more accurately, which provide accurate basis for clinical treatment of cerebral hemorrhage.Finally, by analyzing the marching cubes algorithm, this paper proposes a MC-based fast iso-surface extraction method. According to the vertices relativity of cubes, the proposed method judges discards or reservations of the data points with the suitable angle and reduces the operations with mid-point method instead of the traditional linear interpolation. Compared with existing marching cubes algorithm, this method has several advantages, such as avoiding the topology ambiguity, accelerating the speed of iso-surface extraction, improving the efficiency of 3D reconstruction and enhancing the quality of the reconstruction image. In the end, the experiments demonstrate the feasibility of the proposed method.