Positioning error analysis of a land-based dual-camera mobile mapping system using POS AV 510

Mobile mapping system (MMS) is widely used for fast and large quantify of geospatial data acquisition. The integration of GPS/INS with digital camera or laser scanning for direct geo-referencing is a pre-requisite for MMS. Comparing with conventional in-direct aerial triangulation method, direct geo-referencing technique is more efficient and economic in the derivation of camera's exterior orientation parameters. In this research, we apply the Applanix POS AV 510, which is designed for airborne vehicle MMS, in a land vehicle MMS for the acquisition of facade texture information. The system includes two DSLR digital cameras with 30 degrees tilted-up angle and was installed as looking forward and backward, respectively. The image acquisition was controlled and saved by the notebooks and their trigger events were stored in the POS's memory card concurrently. The error budgets for the direct geo-referencing were evaluated including the surveying of calibration field, lens distortion correction, and boresight/lever system calibration. Preliminary experimental results show that the calibration field is the most critical error source in direct georeferencing. However, due to the calibration field has wide open sky for GPS signal reception and a static data collection scheme is proposed during calibration, the designed system can achieve less than 6 cm of 3D positioning error when single camera is used and a 3cm of accuracy could be achieved when dual camera convergent imaging is adopted. The results are promising for facade texture generation and topographic mapping.