Geometric calibration for stereo near-infrared camera using an 850 nm illuminant array

The advantages of optical surgical navigation systems in surgical applications have been extensively reported in the literature. As the eyes of the system, near-infrared cameras play a crucial role. Currently, the calibration target required for near-infrared cameras is expensive and complex, and there is a lack of high-precision feature extraction methods, resulting in large calibration errors, limiting the improvement of positioning accuracy in navigation systems. In response, a novel and simple near-infrared calibration target with an asymmetrical arrangement of control points is presented, which allows the tedious and time-consuming sorting task to be automated. Additionally, we compare the performance of several feature extraction methods, showing that the proposed Zernike moments-based method yields the best results in terms of mean reprojection error and reconstruction accuracy.