An improved two-stage camera calibration method based on evolution calculation

According to the calibration of binocular vision, an improved two-stages camera calibration method involved with multi-distortion coefficients is introduced in this paper. It takes the sum of distance in the world coordinate between ground truth 3D points and calculated 3D points as cost function which is optimized by evolution algorithm (EA). The 3D pointspsila coordinate are calculated by the imitated direct linear transformation (DLT) triangulation based on distortion compensation. This strategy can allow the two cameraspsila all parameters being optimized at the same time. Parameters optimization algorithms with genetic algorithm (GA) and particle swarm optimization (PSO) are introduced. Simulation and experiment are made under the same calibration data sets. A comparison for GA, PSO and Levenberg-Marquardt (LM) algorithms are made. The results show that the strategy of taking the 3D reconstruction errors as object function is feasible, the evolution algorithms have much higher calibration precision than that of the LM algorithm.