A method of vehicle camera self-calibration

On account of the poor anti-nose performance and the complicated calibration process in the existing algorithms of the vehicle camera self-calibration, a method based on the single octagonal template is proposed in this paper. This method takes advantage of the character of cross ratio and the harmonic conjugate in the projective geometry and the relevant reasoning of Laguerre theory. With this method, the information about the size, locations and motion states of the template isn´t needed while reducing the movement times of the template. Meanwhile, the precision of the corner detection can reach the sub-pixel via improving the method of Harris corner detection. A large number of experiments proved this method high-precision and effectiveness while giving consideration to the noise and other factors in the real conditions.