The Research On ROI Extraction Of3d Volume Data And Its Applicatio

With the continuous development of science and technology, a large number ofcomputer-generated data need to be visualized. The purpose of it contains the secondaryanalysis, extraction of target data and objective reproduction. Volume data visualizationtechnology is an important part of computer graphics. It is a theory method and technique thatcan convert the data to graph or image and displayed on the computer screen, and interactiveprocessing.In the application of volume data, the extraction of the region of interest (ROI) as a partof visualization has a very important practical significance. In this paper, first we introducethe extraction of ROI, and the definition of the volume data and the method of its rendering.Then we analysis and realize two types of region of interest based on slice and isosurface.Base on the library of coin3d. Two kinds of ROI, orthogonal slices and subvolumes areimplemented in this paper. Seismic data is taken as an example to show the results of volumerendering. And the large scale data are loaded and rendered dynamically using the octree.Also we introduce a variety of isosurface extraction algorithm. Such as Marching Cubes andMarching Tetrahedra, etc. The ideas and problems are described in detail, and theimprovements policy are proposed in the paper. The example of the isosurface extraction usesthe famous algorithm called Marching Cubes, and which is improved based on theneighborhood voxel in this paper.There are several problems in the isosurface extraction, such as data noise and blurcontour. The level set method has been widely used in the field of image segmentation. Thelevel set method makes the evolution of the object in n-dimension be converted to n+1dimension. By solving its zero level set, we get the evolution results of the n-dimension. Thismethod has good topology adaptive capacity. The traditional level set method frequently usedin the2D image segmentation. It can identify the object contours in grayscale image. Here weexpand the method to be used in the3D image data. We use the geometric active contourmodel, solving partial differential equations of the energy functional which define with thesource image and the level set function, obtaining iterative form of the level set functionevolution. The level set function evolves with the mathematical and numerical mechanisms, eventually reaches the minimum value of the energy functional, completes the evolution ofthe contours, and obtains the final clear isosurface. This paper uses the simplified model ofMumford-Shah model called C-V model, to achieve the evolution of the level set function.The energy function of the model uses the global information of the image, which producesgood three-dimensional segmentation result. We use the level set method to extract thecontour of the three dimensional data, which can as a data preprocessing procedure of theisosurface extraction. As the level set function using the signed distance, which means theshortest distance between the points and the target contour. If the point is in the target contour,the distance is positive, and negative outside the area. So we can use the feature to determinea narrow band around the level set function, and the data beyond the narrow band can be seenas the noise data. Only the points in the narrow band can be used as the extracted isosurface.This method can control the noise data in the isosurface extraction. The isosurface becomemore clearly and effectively.