The 3D-Reconstruction Of The Man's Skull Based On Serial Stereo Images Of Structured Light

3D Reconstruction of Medicine Image is the important field of research and application in scientific compute visualization. And to realize the 3D Reconstruction of Medicine Image can reappear the outline of the tissue and organ. In the medic diagnose, doctors need some effective tools to rebuild all kinds of man's organs and display them for the purpose of confirming the pathological changes. Nowadays, the most 3D Reconstructions of Medicine Image is based on the X-ray Images, and we get 3D information from the 2D X-ray Images by the theories of computer vision.The paper introduces the photogrammetric methods to reconstruct the man's skull based on serial stereo images of structured light. The paper consists seven chapters and expounds the whole process of the 3D Reconstruction of the man's skull. The main content of the paper is as follow:(1) camera calibrationUse the single-view calibrating method of vanishing points to calibrate the non-metric digital camera.(2) image matching of the object with indigent textureBecause the surface of the man's skull is absent of the texture, we are difficult to match the identical points. The paper introduces a method based on structured light to find identical points on the Epipolar Lines and the laser feature lines. The method resolves the difficulty of the matching of the identical points, and it's easy to be carried out and requires low hardware. So the method the paper protests can be popularized in the 3D Reconstruction of the objects with indigent texture or in the condition of bad illumination.(3) the Relative orientation of the stereo images based on Relative orientation Linear Transform(RLT)For the close range photogrammetry is different from the traditional air photogrammetry and the stereo images' obliquity is big and obliquity's approximation is unknown, we can't process relative orientation by traditional iterative method. The paper uses the Relative orientation Linear Transform(RLT) to process relative orientation and resolves the problem very well.(4)generate TIN (Triangulated Irregular Network) and connect models dividuallyFor the stereo images' big obliquity, we can't meet the need that every image has the same spatial identical points in the large field of vision. So we must separate the skull model to several parts and ensure that in every separate part every image has a few spatial identical points so that we can connect the models.At the same time, when we generate TIN, we must project the spatial points to the plane and every point can't overlap each other. So we also must generate TIN dividually then merge them into the whole skull's DSM.By this dividual generation of TIN and model connection, we predigest the process of thebuilding of DSM and improve the precision of model connection.(5)generate constrained Delaunay triangle meshTake the feature lines and boundary lines as the constrained border, generate the CD-TIN.CD-TIN can express the skull's DSM better.(6)texture mappingTake the original images of skull as the texture images and add the real texture to the skull model by the technology of OpenGL texture mapping.The skull model with real texture are more vivid and stereo.