Accuracy enhancement of UAV products using GNSS derived ground control points and varying flying heights | |
| Author | Naidu, Dharmana Harish |
| Call Number | AIT Thesis no.RS-17-29 |
| Subject(s) | Flying-machines--Remote--Sensing Point |
| Note | Submitted in partial fulfillment of the requirements for the degree of Master Engineering in Remote Sensing and Geographic Information Systems |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis;no. RS-17-29 |
| Abstract | Drones are quickly becoming the go-to means for the collection of on-demand aerial imagery across industries such as construction, surveying, insurance, and mining. Photogrammetry allows us to digitalize the physical world and use that data to solve some of today’s toughest challenges. In the process, it eliminates the need to manually capture data in dangerous areas such as industrial jobsites, quarries, roofs, and other elevated structures. Photogrammetry relies on cameras to measure real-world objects and turn 3D space into 2D maps. This requires photogrammetry software to identify the location, orientation, and movement of the camera to calculate the position of three-dimensional points. Without a physical relationship between the camera and the subject being mapped, it is difficult to quantify the accuracy of photogrammetric outputs. Therefore, to properly quantify error the outputs must be ground-truthed against known values. Cameras are physical devices that introduce errors into the data capture process. A camera’s 2D image isn’t a true representation of the physical world. This is because camera bodies and lenses cannot be manufactured perfectly, which create errors—such as distorted lines in a photo—that photogrammetry software must compensate for.These errors, even when compensated for, can create inaccuracies in linear measurements made on a processed map. In this study we investigate different ways to improve mapping accuracy and put together a set of best practices to be used when making linear measurements. To test the accuracy of the measurements made using maps generated at different heights from data captured with industry-standard DJI drone platforms, we established 33 ground control points using Topcon GR-5 GNSS receivers and distributed equally in our project area. Then, more than 25 flights were logged—each exploring different flight altitudes, cameras, and photo overlap settings. The images collected were then processed in the Pix 4D Pro and Agisoft Photoscan Professional using ground control points and without using ground control points. The data sets were analyzed and used to calculate the average margin of error for measurements of known control points. Using the results, we determined that using a performance camera and flying at low altitude with high image overlap produced maps with the best linear measurement accuracy. We also determined that as height increases ground control points must be used to georeferenced the orthoimage and get good linear accuracies. |
| Year | 2017 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis;no. RS-17-29 |
| Type | Thesis |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing (RS) |
| Chairperson(s) | Tripathi, Nitin Kumar |
| Examination Committee(s) | Apichon Witayangkurn;Pal, Indrajit |
| Scholarship Donor(s) | AIT Fellowship |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2017 |