Free-form surface reconstruction from 3D multi-resolution data on spherical space

Over the last two decades, the 3D scanning technology has evolved steadily and facilitated the reconstruction of complex engineering parts. Using a 3D laser scanner, a physical object can be copied and stored in the form of a computer model. Based on the sampled points on the object's surface, called `points cloud', the object can be reconstructed via triangulation and surface meshing to a 3D solid model. There are various surface reconstruction methods exist. Unfortunately, most of those methods require the reconstruction process should always begin from the 2-dimensional Euclidean (ℝ²) space. This condition limits the ability for fast reconstruction. This paper introduces a new surface reconstruction approach. The proposed method transforms the vertices into spherical polar coordinates instead of Euclidean coordinates, and triangulation begins directly on the S space. Thus, the whole process works in a faster and more efficient fashion. The proposed algorithm has been implemented using C++ in combination with OpenGL and Qt 4.7 graphic libraries, and a test-bed is performed for verification.