Modeling And Rendering Of Voxel And Their Applications In Medical Images

olume data is a set S of samples (x, y, z, v), representing the value v of some property of the data, at a 3D location (x, y, z). Volume data are obtained by sampling, simulation, or modeling techniques. Analyse of volume data is concerned with volume data representation, modeling, manipulation, and rendering for extracting meaningful information from volume data and supporting the posterior CAE. For example, a sequence of 2D slices obtained from Magnetic Resonance Imaging (MRI) or Computed Tomography (CT) is 3D reconstructed into a volume model and visualized for diagnostic purposes or for planning of treatment or surgery. Industrial CT is used for non-destructive inspection of composite materials or for building CAD models in reverse engineering. Similarly, confocal microscopes produce data which is visualized to study the morphology of biological structures, etc. The main contents of this dissertation include voxel modeling, visualization and their applications in medical images. An object is represented with its voxel surfaces in voxel modeling, which uses a point in 3D space (voxel), rather than a polygon, as the surface primitive.Volume data segmentation is the premise of voxel modeling. In this dissertation the method of segmentation mainly focused on medical image is discussed. Two methods to segment medical images are given: One is the medical image segmentation based on gray-level thresholding, the other is using neighborhood operation for segmentation of tiny bone from computer-aided tomography images.The definition of the voxel surfaces of an object is proposed in this dissertation. It is proved that the voxel surfaces of an object are closed and 18-connected. The surface data is stored in a structure that consists of two parts, a Surface Voxel List (SVL), and an Index Table (T). Two methods of estimating the surface normal are provided: grey-level gradient shading anddistance gradient shading.Because the surface voxel can be rendered as a point, rendering of voxel surface is very efficient. But clue to its discrete form, voxel surface may yield high aliasing artifacts during rendering. The resampling process is used to simulate the process of the image transform and the reasons of why aliasing artifacts are yielded are given by using sampling theorem. Two anti-aliasing methods for voxel surface rendering are presented: one is to average the normalized normals of the surface voxel and its adjacent voxels weightedly, the other is to create a rendering surface and use it to display instead of voxel surface. To improve the smoothness of the rendering surface, a smoothing operation of the normalized normals on the object surface produced by Dividing Cubes algorithm is presented. The aliasing artifacts that ray sampling may yield in perspective projection is discussed, too. Mosaic may appear when zooming in on the voxel surface. In this dissertation an interpolation magnification algorithm based on the dividing curves is proposed, which is used to magnify the projection normal image of the voxel surface. Then the algorithm is generalized to magnify gray images and the dividing curves are fitted by using cubic uniform B-splines. The result of the magnified gray image not only preserves sharp contours but also achieves smoothing image.At the final of this dissertation an application program interface (API) pocket to implement the voxel modeling and its relative techniques is introduced.