A high-precision measurement system with a large field of view

The sources of measurement errors of stereo vision systems include model approximation errors, data errors, and calibration errors. Due to the existence of these errors, the commonly used binocular self-calibration method is difficult to satisfy the measurement system that pursues high precision. And because the comprehensive index of the size of the field of view and the absolute measurement accuracy is the key to evaluating the stereo vision system. Therefore, this paper designs a vision system with a large field of view and high precision, which is composed of a static surveillance camera and an active gaze camera. And for this system, a joint calibration scheme based on error correction matrix is proposed, which only contains one geometric constraint of ideal rotation axis. Since the nonlinear system of equations formed by these matrices cannot be solved analytically, a numerical optimization solution based on Lie algebra is proposed. Specifically, matrix acquisition is formulated as the problem of minimizing the sum of distance deviations between ideal predicted values and actual observed values. This problem is then solved by differentiating these matrices in the form of left-multiplying tiny perturbations. In this way, a high-precision joint calibration with multi-source errors is achieved. Finally, the effectiveness and robustness of the proposed method are verified by experiments.