Three-dimensional Reconstruction Of The Medical Ct Images

With the progressive development in science and technology, people have improved their ability to understand themselves with the applied science. After the finding of X-ray, some advanced medical devices such as CT, MRI and B-Ultrasonic come into appearance. This made the traditional medical diagnosing changed much. With the quick development of computers, doctors performs analyzing and processing on medical images with computers. Nowdays, the 3d reconstruction of medical images has become a research hot spot.There are two methods in 3d reconstruction of medical images: volume rendering and surface rendering . Because surface rendering can get acceleration with current hardwares, it can render with high speed and is suitable in run-time interaction. So it is widely used. Currently, Cuberille algorithm , Dividing Cubes algorithm, Marching Cubes algorithm are main surface rendering algorithms in use.The Marching Cube algorithm is the classical iso-surface extracting algorithm in regular volume data. The MC algorithm scans all data units in 3d space one after one. But according to some literatures, the cubes intersecting with the iso-surface only occupied quite a little of the whole data, and this make the mc algorithm have only low efficiency. In this thesis, a seed-occupying algorithm is proposed to improve this disadvantage. The basis of the new algorithm is that most iso-surface is continuous. Firstly find a part of the iso-surface; set that part as a seed, then expand the seed to the whole space and find the complete iso-surface. The new algorithm expand the iso-surface directly, discarding the one-by-one scanning. This decrease the additional scanning of the empty unit, and improves the efficiency highly.After constructing an interpolating surface on the iso-surface mesh, the iso-surface can be got. To preserve the smoothness on the iso-surface, we must assure the c1 continuity, i.e. the normal-continuity on the iso-surface. In fact, after defining the normal on the iso-surface mesh, the normals on other places can be acquired by...