An approach to improve direct sensor orientation using the integration of photogrammetric and lidar datasets

Abstract

ABSTRACTThe Direct Sensor Georeferencing (DSG) can co-register both photogrammetric and light detection and ranging (lidar) datasets in a common mapping reference system. This procedure is based on the information (position and orientation) derived from Global Navigation Satellite System (GNSS) and Inertial Measurement Unit (IMU) data. Due to inaccurate system mounting parameters, the attitudes and positions are badly calculated. In this scenario, the accuracy of the boresight misalignment angles (physical mounting angles between digital camera and IMU) is critical to obtain models free of y-parallaxes. Therefore, a system calibration procedure is required to improve DSG using photogrammetric and lidar datasets. In this sense, this article aims to develop an approach to improve DSG based on the enhancements of the accuracies of the photogrammetric and lidar datasets integration. For this purpose, one sub-block was extracted at the centre of the entire block. Then, the in situ camera calibration was performed using the sub-block and sets of Lidar Control Points. Sequentially, a boresight calibration was performed using Virtual Lidar Control Points. The quality assessment of the approach was measured in both image and object spaces, using two-ray checkpoints at the entire block. The overall results from the performed experiments showed significant improvements of the Root Mean Square Error of the checkpoint discrepancies computed by direct georeferencing.