A new method for camera self-calibration based on the linear iteration is presented in the paper.Without any prior-information about the scene provided

the self-calibration is performed by the linear iteration.The iterative algorithm is executed in two steps:Firstly

assuming the skew

s

=0 and the principal point lies in the origin

the absolute quadric is acquired in the projective space and then the camera intrinsic matrix is found.Secondly

with the result of the first step

the image coordinates are translated to satisfy the image captured by the camera hypothesized in the first step.These two steps are repeated and the real camera intrinsic matrix is obtained finally.The experiments with both simulate and real data show that the self-calibration method is efficient

robust and have good property of convergence.