Quantitative visualisation of surfaces from volumetric data

Jonathan C. Carr1, Andrew H. Gee1, Richard W. Prager1 & Kevin J. Dalton2 1Dept. of Engineering, University of Cambridge, Cambridge, UK. Email: [email protected] 2Dept. Obstetrics & Gynaecology, University of Cambridge, Cambridge, UK. ABSTRACT A 3D B-spline quasi-interpolant is used to extract smooth iso-value surfaces from volume data. In this technique discretization and partial voluming artifacts are reduced by approximating sampled data at voxel centres. Surface normals, necessary for realistic shading, are analytically de ned by the approximating function rather than estimated by in an ad hoc way from the volume data. We consider an application where bone surfaces are revealed from CT data by ray-casting and the surfaces are then used to construct models and prostheses. Accurate determination and rendering of bone surfaces is required. A z-bu er shading technique is also described for improved rendering of surface depth-maps.