Research of camera calibration algorithm in the auto rack girders detecting system

For online accurate holes detection of the auto rack girders, this paper proposes a method of combination non-linear distortion correction and double-linear transformation, to calibrate internal parameters and correct geometric parameters. Firstly, in different angles, through 20 grabbing calibration templates images, carrying on non-linear distortion correction for grabbing images with Zhang Zhengyou calibration algorithm, to gain internal radial and tangent direction distortion parameters. Through experiment data, we find internal parameters want to tend to be stable only through more than 12 calibration templates images; secondly, after canceling lens distortion, because included angle exists between imaging plane of camera and under-detecting plane the auto rack girders, circular hole may become elliptical after grabbing images. So we images are carried on a double-linear transformation again, which have been carried with non-linear transformation, namely let imaging plane of camera becomes parallel to the auto rack girder plane. Then at this point, calculation of image may draw near true value of measurement. In this paper, we discussed the possible deviation when camera calibrates. Experimental results indicate calibration algorithm is practicable for holes on the auto rack girders, which is about combination non-linear distortion correction and double-linear transformation, its calibration accuracy is 0.41 pixels.