A linear iterative least-squares method for estimating the fundamental matrix

During the last two decades a lot of researches have been done on the estimation of fundamental matrix, which represents the epipolar geometry between two uncalibrated perspective images. In this paper, a new linear and iterative method is proposed for estimating the fundamental matrix. It preserves the noise model of the observed image points, e.g. a Gaussian noise distribution. When the noise in the measurement of the image points is small, the accuracy of this method is comparable to that of the nonlinear Newton-type optimizers, however, it is much more efficient both because of its linearity and because of its faster convergency.