A linear solving method for rank 2 fundamental matrix of non-compulsory constraint

Solving the fundamental matrix is an important research topic in computer vision. The relationship between the epipole and the parameters of fundamental matrix can be found from the fundamental matrix of rank 2. A new model is equivalent to the fundamental matrix of rank 2. The model of the fundamental matrix, whose rank equals 2 can be provided. According to the relationship between the parameters, the epipole and the fundamental matrix model, a linear method which avoids the objective function of unconstraint programming solving a nonlinear equation with the element 4 and the power 8 is provided. It realized stable estimation of the fundamental matrix. In the same scene, our algorithm compared with the 8-Points algorithm and the RANSAC algorithm, indicates that our algorithm has smaller errors under certain case. The comparison of results indicates our method algorithm is feasible and has stronger practicability by experiment.