Abstract: The paper presents a robust self-calibration algorithm based on iterative weight, which can efficiently discard the outliers. The weight of each point is determined based on the re-projective errors and the metric reconstruction is obtained based on the weight. After serveral iterations, the weights of the outliers approaches zero, with the projective reconstruction obtained with good accuracy. The camera intrinsic parameters are obtained after projective reconstruction. The algorithm can overcome the drawbacks of both the least-squares method and the RANSAC method. The theory and experiments with both simulation and real data demonstrate that the self-calibration algorithm is very efficient and robust.

Key words: self-calibration, re-projection error, iterative weight