Infrared camera arrays have previously demonstrated the potential for long-range detection due to their distributed aperture structures, but their performance is affected by the calibration procedure. The widely used calibration method based on image registration estimates the transformation matrix between different cameras by extracting matching features, and its accuracy is limited by low-resolution infrared images. To solve this problem, a new calibration method based on infinite scene registration is developed, which has high registration accuracy and is easy to implement. The results of the laboratory experiment demonstrate that the signal-to-noise ratio of the point target is improved by 1.86 with the proposed method, which proves the effectiveness and competitiveness of the proposed method. Finally, the measured data of long-range aircraft further verifies the feasibility of the proposed method.