A Performance Evaluation of Using Fish-Eye Lenses in Low- Altitude UAV Mapping Applications

One of the major uses of unmanned aerial vehicles (UAVs) has been to provide aerial imagery for applications, such as quantitative remote sensing and surveillance. UAVs have the potential to provide this imagery in a much more cost-effective manner than satellite systems. Additionally they have the added advantage of being able to collect the data on an “as needed” basis (eg. daily) and can also fly below cloud cover. However, there are technical challenges such as blur effects, payload limits and the lower altitudes of operation which reduce the swath width. The aim of this research is to understand the benefits and limitations of using a downward looking camera with a fish-eye lens for low-altitude UAV aerial mapping missions. Fish-eyes have been found to have valuable characteristics in their ability to capture scene data covering a wide field of view (FOV). Lens makers have copied this characteristic of a large FOV and applied it to fish-eye lenses, which are commonly used in photograph. The fish-eye lens has potential benefits in UAV terrain mapping, particularly at low altitudes, because of its large FOV. The large FOV makes the camera less sensitive to movements of the aircraft. A downward-looking camera provides images that show a 180x360 degrees view of the scene, which also includes the horizon. The ability of the downward looking camera to capture details of scene in front and back at the same time provides additional information that can be used for height estimation of ground obstacles and attitude estimation of the aircraft. However, the fish-eye lens also produces heavy distortion in the captured images, which needs to be rectified. The lower altitude of operation produces a larger motion blur component which must be deblurred. The rectified and deblurred images will build the basis for the mapping process, which merges single, distorted and rotated pictures to a mosaic map. This paper presents the benefits and limitations of using fish-eye lenses in low-altitude UAV mapping applications. The mapping uses only lightly distorted parts of the fish-eye images. Further, techniques to restore uncovered areas using the heavily distorted parts of the fish-eye images are introduced. Finally, this paper demonstrates precise processes to rectify the distorted fish-eye images, addresses deblurring processes and some quality issues for quantitative remote sensing purposes.