Research On Camera Calibration Algorithm In Material Pile’s Volume Measurement

For the effective measurement of the large material pile in enterprises, this paper explores a camera calibration algorithm about the volume measurement of the material pile, which is based on machine vision, to ensure the accuracy of the camera calibration, and what all these do is to bring about the accurate measurement of the volume for the material pile.The main contents of this paper are listed as follows:(1) Design the calibration board and test board with a black and white grid (25mm X 25mm), then complete the device of the whole experiment measurement system, enabling the camera to be installed to the pan-and-tilt stabilitily and ensuring the camera’s 3D position..(2) Analysis the causes of the distortion in the process of the camera imaging, then use the model that exists the radial distortion for the camera calibration.(3) Use Tsai two-step method for camera calibration that based on the radial constraint. First of all, find out the rotation matrix, tx, ty of translation matrix, and the scale factor of the image in the process of calibration. Then get the effective focal length f, t. of translation matrix and lens distortion coefficient. In the second step of solving the problem, this paper chooses the particle swarm algorithm to optimize these parameters of the camera, in order to make the camera calibration parameters to be solved.(4) Do experiments with the test board and recover the world coordinates by the way of solving the parameters, then compare with known real coordinates of test on board. In the process of recovery, the single point calibration fiducial error value< 1.06%, and the repeatability error< 5.43% in the X direction; The fiducial error value< 1.90% and the repeatability error< 4.03% in the Y direction; The fiducial error value< 3.74% and the repeatability error< 13.89% in the Z direction.(5) Do deflection correction for the 3D point of the material pile, using interpolation by the algorithm of radial basis interpolation function for the corrected material pile scatter points. This experiment achieves 3D reconstruction and the volume calculation of the material pile. It also assures the relative error of the material pile volume ranges from 2.78% to 7.84%.This topic achieves the camera calibration, reaching the material pile measurement accuracy through modeling tests in lab. The method is simple and low cost.