Calibration of boresight offset of LROC NAC imagery for precision lunar topographic mapping

High-resolution and high-precision lunar topographic information is essential for lunar exploration and scientific research. The lunar surface imagery acquired by the Narrow Angle Cameras (NACs) of NASA's Lunar Reconnaissance Orbiter Camera (LROC) is a valuable dataset for lunar topographic mapping because of its very fine ground sampling distance (GSD) of 0.5-2 m. The NACs have a unique configuration consisting of two separate cameras closely mounted. This configuration enables the collection of lunar surface images with both large coverage and high resolution. However, the mapping results from the NAC images depend highly on the accuracy of the relative alignment (referred to as boresight offset) between the two NACs. This paper firstly presents the details of the NAC sensor configuration and orientations. Then, an approach to calibrate the boresight offsets of the two NACs is presented. By using triple-matching tie points between the NAC images, the boresight offsets are improved based on the geometric model through a least-squares adjustment. Experiments using typical NAC stereo images show that there are inconsistencies as great as 80 m in object space in the topographic models generated using the pre-flight determined SPICE kernels. The inconsistencies are well improved using the image orientation parameters derived from the temperature-dependent SPICE kernels recently released by the LROC team, however there are still small inconsistencies of about 5-10 m in object space, which can be further reduced to meter level by using the method proposed in this paper. This method is of significance for generating lunar topographic models with high precision and internal consistency from NAC imagery.