A novel calibration method for a large-scale stereo vision sensor based on a one-dimensional target

Abstract

Large-scale stereo vision sensor is of great importance in the measurement for large free-form surface. The intrinsic parameters of cameras and the structure parameters of the stereo vision sensor should be calibrated beforehand. Traditional methods are mainly based on planar and 3D targets which are expensive and difficult to manufacture, especially for large dimension ones. A calibration method for stereo vision sensor based on one-dimensional targets is proposed. First random place two 1D targets, and acquire multiple images of the targets from different angles of view with camera. Solve the intrinsic parameters of camera with the constraint that the spatial angle of the two ID target are constant. Then set up the stereo vision sensor with two calibrated cameras, and acquire multiple images of a 1D target of unknown motion. Based on the constraint of the known distance between two feature points on the target, estimate the initial value of the structure parameters with linear method and the precise structure parameters of stereo vision sensor with non-linear optimization method by setting up the minimizing function involving the scale factors. Experimental results show that, the measurement precision of the stereo vision sensor is 0.052mm, with the working distance of 3500mm and the measurement scale of 4000mm × 3000mm. The method proposed is proved to be suitable for field calibration of stereo vision sensor in application of large-scale measurement for its easy operation and high efficiency.