Aerial Pose Detection of 3-D Objects Using Hemispherical Harmonics

In this paper, we consider pose estimation of 3-D objects from an aerial perspective using eigendecomposition. We first outline a sampling method to acquire images of the object from an aerial view by sampling the upper unit hemisphere. Using this hemispherical sampling pattern, the computational burden of computing the eigendecomposition can be reduced by using the Hemispherical Harmonic Transform (HSHT) to "condense" information due to the hemispherical correlation. We then propose a computationally efficient algorithm for approximating the eigendecomposition based on the HSHT analysis.