Combined calibration method for large field of view multi-camera system

In the calibration process of multi-camera system with large field of view, aiming at the problem of self occlusion in the plane calibration method and the problem of low calibration accuracy of one-dimensional calibration methods, a calibration method of multi-camera system based on the combination of plane target and one-dimensional target is proposed. Firstly, a mathematical model of multi-camera calibration is established with pin-hole model and rigid body transformation theory; the internal parameters and distortion coefficients of each camera are calibrated by two-dimensional plane target; on this basis, analyze the uncertainty of imaging error of circular feature points, cost function of random sample consensus algorithm(RANSAC) is improved. Using the camera imaging model, the essential matrix between cameras is estimated, and the initial value of camera external parameters is decomposed and determined; lastly, multiple nonlinear optimization algorithms are used to refine and obtain the camera matrix, The simulation and actual experimental results show that the multi camera system combination calibration method proposed in this paper can realize the parameter calibration of four camera measurement system, and the measurement space is 2000mm×2000mm×2000mm, the 3D reconstruction error is less than 0.15mm. Compared with the traditional calibration method, it has more accurate calibration accuracy.