Large spacecraft and satellites can suffer costly on-orbit failures. In many cases, it would be helpful to have a small inspection satellite that can detach from and visually inspect the larger host spacecraft. This type of small inspection satellite must be carefully designed to minimize both the requirements imposed on the host and the risk of collision. For this reason a relative navigation system must be highly reliable and accurate. This paper discusses the design, implementation and testing of a navigation system that utilizes singlecamera computer vision based navigation to estimate the 12 Degree of Freedom relative state. This system uses a fiducial target that is designed to be as small as possible, in order to reduce the space requirement on the host spacecraft. The difficulties of using only the minimum point features are discussed and a nonlinear estimation approach is presented that can overcome these issues. These algorithms are implemented and experimentally tested on the SPHERES satellites, and an upper bound on the precision and accuracy is reported.