Location: Range Management ResearchTitle: Orthorectification, mosaicking, and analysis of sub-decimeter resolution UAV imagery for rangeland monitoring Author
|Rango, Albert - Al|
Submitted to: Society for Range Management Meeting Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 9/1/2010
Publication Date: 2/6/2011
Citation: Laliberte, A.S., Winters, C.D., Rango, A. 2011. Orthorectification, mosaicking, and analysis of sub-decimeter resolution UAV imagery for rangeland monitoring [abstract]. 64th Annual Society for Range Management Meeting. p. 137. Interpretive Summary:
Technical Abstract: Unmanned aerial vehicles (UAVs) offer an attractive platform for acquiring imagery for rangeland monitoring. UAVs can be deployed quickly and repeatedly, and they can obtain sub-decimeter resolution imagery at lower image acquisition costs than with piloted aircraft. Low flying heights result in imagery highly suitable for mapping soil and vegetation types, structure, and pattern in great detail. Small UAVS are commonly equipped with lightweight digital cameras due to low payload capabilities, resulting in challenges associated with photogrammetric processing and creation of orthomosaics from large number of small footprint images. We developed a custom, semi-automated approach that is suitable for processing hundreds of UAV images into orthorectified image mosaics. A customized algorithm improves the accuracy of the UAV’s exterior orientation data, comprised of position (X, Y, Z) and attitude (roll, pitch, heading) information derived from the UAV’s flight computer. The corrected exterior orientation data are subsequently used as inputs for orthorectification and mosaicking with minimal or no need for tie- and/or ground control points, greatly reducing time and cost of processing. The workflow has been tested on 65 image mosaics of arid rangelands with few distinguishing features. Orthomosaics created using this process have positional accuracies of 1 m in flat terrain and 1.9 m in hilly terrain. Object-based image analysis of the image mosaics has resulted in classification accuracies of 78%–92%, depending on vegetation type and number of classes. The results show that UAVs are viable remote sensing platforms and that quality products can be derived from the imagery.