A mobile stereo vision system with variable baseline distance for three-dimensional coordinate measurement in large FOV

A novel mobile stereo vision system (MSVS) with variable baseline distance for three-dimensional (3D) coordinate measurement in large field of view (FOV) is established herein. Each independent mobile camera can rotate in the horizontal and vertical directions, and the position of the cameras are obtained by differential GPS in real time. In order to achieve rapid camera calibration in large FOV, a simplified model of MSVS based on differential GPS is constructed in this paper. A six-point method is proposed to quickly estimate the camera's initial parameters (i.e. focal length, initial roll, pitch, and yaw angles) in the situation that the approximate value of the initial yaw angle is unknown. Moreover, the camera's roll angle is equivalent to its approximate value measured by a high-precision inclinometer during the movement, and the unknown extrinsic parameters of each camera are reduced to the pitch and yaw angles after the camera moves. The refined pitch and yaw angles are eventually estimated by only using a single control point, which makes it possible that the 3D coordinate can be measured online after the camera moves. The computer simulation verifies the effectiveness of the extrinsic parameters calibration method using a single control point. The quantitative results demonstrate that the standard deviation of the focal length of the camera does not exceed 0.028 mm, and the mobile camera's pitch angle and yaw angle estimated by a single control point is extremely close to the reference values. In addition, the average error of the 3D coordinate on each axis after the camera moves is smaller than 0.08 m in the measuring distance of 100 m, and that on X-axis and Z-axis is comparable and does not exceed 0.35 m. The proposed methods are suitable for the occasions that high accuracy is not required in the field of 3D coordinate measurement in large FOV.