| An estimate of the epipolar geometry is vital in applications such as three dimensional reconstruction of a scene. Epipolar geometry is the intrinsic projective geometry between two views and depends only on the cameras' internal parameters and relative pose. It is characterized by a 3 x 3 matrix called the fundamental matrix. This thesis endeavors to replace the classical epipolar estimate, which is based on image feature correspondences, with one based on a homography. This will allow the use of a less sophisticated algorithm to calculate the fundamental matrix because an area-based correspondence is more robust to noise. The proposed algorithm is an amalgamation of the Video Orbits technique and the normalized eight-point algorithm. The results are very robust to image noise. Due to the fact that the estimate uses a linear algorithm, it quickly delivers a result. However, this comes at the cost of not being as good as the gold standard. |
