<p>The calibration of a camera array with a nonoverlapping field of view is a fundamental task in advanced industrial manufacturing production. This task involves the development of transformation relationships between cameras without a shared field of view. In this study, we propose a circular coded ring target pattern for nonoverlapping observations to provide reliable control points. Precision positioning is addressed by utilizing the multiscale circle characteristic of the pattern and the gradient centroid method to extract the positioning center. The encoding principle ensures the uniqueness of homonymous points. Note that the cameras only require partial observation of the calibration board to deduce their relative positions based on virtual symmetric transfer errors. Additionally, we present an expandable calibration model that is customized for maneuverability in challenging environments. Finally, the reprojection error and geometric constraints are introduced for optimization. We analyze the proposed method in synthetic and practical industrial production applications and achieve good results. This shows that the method is effective and can be adapted for various camera configurations.</p>