Image interpolation is one of the most important steps in three-dimensional reconstruction which can offer real 3D scene for clinical diagnosis and therapy. By applying interpolation on slices, anisotropic data can be changed into isotropic one, which could highly improve 3D reconstruction's standard. This paper introduced several interpolation methods, which were further classified as scene-based interpolation and object-based interpolation.Scene-based interpolation based on sinc interpolation was first introduced which includes polynomial, Lagrange, B-spline and windowed sinc interpolation. The nearest neighbor and linear interpolation also belongs to this category.Next, object-based interpolation was introduced which includes match interpolation, shape-based interpolation and non-rigid registration based interpolation. During this section, some scattered data interpolation methods were also showed up. The first and third interpolations were modified to expecting more reasonable results.Finally, classical ray tracing algorithms was used to rendering the 3D volume data produced by above mentioned image interpolation methods.
|