Real-Time Georeferencing of Image Sequence Acquired by a UAV Multi-Sensor System

Real-time image georeferencing technologies are required to generate spatial information promptly from the image sequences acquired by an airborne multi-sensor system. We thus develop a sequential aerial triangulation (AT) algorithm based on the combined adjustment model. The proposed algorithm enables us to perform AT rapidly with a minimum of computation at the current stage using the computational results from the previous stage during the acquisition of an image sequence whenever a new image is added. The experimental results based on a simulated data set show that the georeferencing of an image sequence at any stage took less than 0.1 second and its accuracy was determined within ±4cm on the ground points coordinates comparing to the results of the existing simultaneous AT. Consequently, this algorithm can be effectively used for diverse applications requiring real-time image georeferencing such as disaster monitoring and image-based navigation.