Research On Camera Calibration Technology For Wheel Alignment Based On Machine Vision

Machine vision has achieved increasingly widespread applications in several fields, such as production testing, biomedicine, military affairs, etc. Compared with traditional mechanical measurement, the method of wheel alignment parameter detection based on machine vision has advantages such as non-contact, convenient measurement, high speed, accuracy, easy maintenance, cost-effective performance, etc. Camera calibration technology is the premise and basis of obtaining 3D information in machine vision, and plays a critical role in the whole system.This thesis researches the key problems in accordance with the process of camera calibration around the objective of high precision by making an analysis and summary on camera calibration technology based on wheel alignment parameter detection system. Adopt the traditional method of camera calibration from a plane based on circular pattern to acquire the intrinsic and extrinsic parameters of camera. Firstly, in the linear camera imaging model, research the deviation of the center of a circle in perspective projection transformation. Based on the theory of perspective projection and theory of spatial analytic geometry, the concentric circular is used to obtain the true projective position of the circular target quickly and accurately. Secondly, research the problem of matching circular markers, and design target pattern which can auto-match markers. Using the topological structure of the four big circles, the circular markers can be detected and matched correctly, so that manual intervention can be shook off in the process of calibration. Then, under the linear camera imaging model, the homography matrix is calculated to acquire partial intrinsic parameters and extrinsic parameters of camera. And in view of the lens distortion, the distortion parameters are obtained based on the concentric circular structure. Finally, the PSO algorithm is adopted to optimize the intrinsic parameters of the camera. Based on the introduces of standard PSO algorithm parameters setting, good convergence is received by better choosing the inertia weigh and the acceleration constants with some factors taken into consideration, for example, the group size.An experiment of camera fixed and shooting target plate (a plane) for getting a number of image data based on monocular vision is designed. The camera calibration method based on matching circular markers and the method based on the conic are compared. The result shows the camera calibration method based on matching circular markers has the higher accuracy and better value of application.