Estimating the fundamental matrix based on the actual distance

The estimation of fundamental matrix is a key problem in computer vision and also the basis of active vision. A new linear iterative algorithm for estimating the fundamental matrix has been presented. The algorithm is based on minimizing the actual distances between the projection points and their matched epipolar lines. The first two fundamental matrixes are computed with the information of images from two cameras, respectively. The final rank-2 fundamental matrix is gained by a certain combination of these two matrixes. The results of experiment with real images show that the algorithm improves the precision and robustness.