A vertex merging algorithm for extracting a variable-resolution isosurface from volume data

This paper proposes a new surface modeling approach, called the vertex merging algorithm, which extracts an isosurface from volume data such as medical images. The marching cubes algorithm (MC), one of the most famous surface extraction methods, generates several million triangles for some medical images. The enormous number of triangles is not suitable for interactive manipulation in terms of both space and time. Hence, reducing the numbers of triangles while minimizing the loss of object details is important for achieving interactive manipulation and saving the space of storage. In order to reduce the number of triangles, the authors introduce a merging scheme which merges the vertices of triangles generated by the MC at a surface generation step. For merging, the angle between two vertex normals of triangles is used for checking whether they should be merged or not. If the angle is less than the angle threshold (AT), two vertices are merged. Using this method, a variable resolution isosurface generation is available if the authors specify a various value of the AT. The experimental result shows that the authors´ algorithm reduces the number of triangles by 52-68% compared with that of MC for some medical images. In spite of this reduction, the quality of the resulting images is comparable with those of the MC