Noise analysis and external parameters calibration for binocular fisheye cameras

Since fisheye lens has a large field of view, it can be applied in principle to positioning in a wide range and a large space. Based on fisheye camera´s internal parameters known a prior, the principle of calibrating the fisheye camera´s external parameters is proposed first. Then the noise distribution of the fish-eye camera image points coordinates in the unit sphere corresponding to the image point coordinates and the impact on essential matrix solution are analysed. Furthermore, it is pointed out that these points with a relatively small incidence angle result in more high-precision external parameters. The simulation results indicate that the analysis on the noise transmission and the impact on the essential matrix solution are reasonable and effective.