Abstract: The method of improving the constraints was proposed to enhance the efficiency of image
rectification and reduce the large amount of calculation caused by using the traditional spherical
perspective projection and intricacy of calibration. Firstly, identify the principal point of the fish-eye
image and its radius by applying Hough transform circle detection to the characteristics of fish-eye
image, and set the condition that a straight line comes out from the same arc through spherical
perspective positive rejection as a constraint. The main view of fish-eye image could be rectified with
the optimal projection sphere radius. Based on this method, a new two-dimension calibration method
is put forward as sector mapping. Start from the trajectory projected from the edge of main view. Map
the fish-eye image in polar coordinates and obtain the rectified surrounding view. Validate the
effectiveness and feasibility of the method by two examples.

Key words: spherical perspective projection, sector mapping, evaluation index, distortion