An Analysis of Sphere Tessellations for Pose Estimation of 3-D Objects Using Spherically Correlated Images

Eigendecomposition is a common technique used for pose detection of three-dimensional (3-D) objects from two- dimensional (2-D) images. It has been shown in previous work that the eigendecomposition can be estimated using spherical sampling in conjunction with the Spherical Harmonic Transform. The issue then becomes deciding on the best tessellation of the sphere to define the sampling pattern. In this paper we evaluate three popular tessellations and compare and contrast their computational performance, as well as their estimation accuracy for the eigendecomposition of this spherical data set.