Geodetic Alignment of Aerial Video Frames

With the sophistication of artificial vision systems, the need to endanger human lives for many hazardous activities is increasingly proving to be avoidable. From aerial reconnaissance missions to space exploration, many projects stand to benefit, in particular, from the sophistication in techniques to precisely find world positions of objects present in video data. Unfortunately, mechanical automation of such a task is complicated by the narrow fields of view of video data and the inaccuracy of mechanical information available describing the position of the camera in the world. Instead, computer vision techniques can be used to successfully align any given video frame with pre-calibrated reference imagery. After alignment, a video frame inherits pixel-wise calibration and as a consequence objects in the frame are exactly placed in the world. This ability to accurately position objects like buildings, roads, landing sites and spatial landmarks in general, facilitates precise automation of actions that previously required human intervention. The core challenge