Templated marching cubes — A low computation approach to surface rendering

Abstract

Surface generation from high resolution datasets using triangulation algorithms like marching cubes require large amounts of computational time for the generation and interpolation of vertices. Here we propose a templated method of generating triangles which has far less computation involved and saves on CPU cycles and memory. Each cube orientation corresponding to the boundary cases in the original algorithm is listed in a prebuilt table of templated triangles. The template created using binary input may be further smoothened using cost functions related to input image data.