Error analysis of a stereo-vision-based tube measurement system

To ensure the accuracy of tube measurement systems based on multi-stereo vision, we defined error sources and their modes of transmission through the system, and developed an error propagation model. In this paper, we first discuss system processes including camera calibration, image processing, point-to-point matching, three-dimensional reconstruction, and error sources identification. Next, we develop a theoretical error calculation equation based on our proposed error propagation model. Experimental verification results showed a practical measurement error of 0.17 mm; the error associated with our proposed theoretical error calculations was 0.192 mm, verifying the utility of the model. The proposed model is expected to help researchers rapidly establish accurate measurement systems that are capable of accommodating processing challenges. (C) 2020 Published by Elsevier Ltd.