Using Geometric Algebra for Optical Motion Capture

Optical motion capture refers to the process by which accurate 3D data from a moving subject is reconstructed from the images in two or more cameras. In order to achieve this reconstruction it is necessary to know how the cameras are placed relative to each other, the internal characteristics of each camera and the matching points in each image. The goal is to carry out this process as automatically as possible. In this paper we will outline a series of calibration techniques which use all of the available data simultaneously and produce accurate reconstructions with no complicated calibration equipment or procedures. These techniques rely on the use of geometric algebra and the ability therein to differentiate with respect to multivectors and linear functions.