A method of fast mosaic for massive UAV images
2014
With the development of UAV technology, UAVs are used widely in multiple fields such as agriculture, forest
protection, mineral exploration, natural disaster management and surveillances of public security events. In contrast of
traditional manned aerial remote sensing platforms, UAVs are cheaper and more flexible to use. So users can obtain
massive image data with UAVs, but this requires a lot of time to process the image data, for example, Pix4UAV need
approximately 10 hours to process 1000 images in a high performance PC. But disaster management and many other
fields require quick respond which is hard to realize with massive image data. Aiming at improving the disadvantage of
high time consumption and manual interaction, in this article a solution of fast UAV image stitching is raised. GPS and
POS data are used to pre-process the original images from UAV, belts and relation between belts and images are
recognized automatically by the program, in the same time useless images are picked out. This can boost the progress of
finding match points between images. Levenberg-Marquard algorithm is improved so that parallel computing can be
applied to shorten the time of global optimization notably. Besides traditional mosaic result, it can also generate superoverlay
result for Google Earth, which can provide a fast and easy way to show the result data. In order to verify the
feasibility of this method, a fast mosaic system of massive UAV images is developed, which is fully automated and no
manual interaction is needed after original images and GPS data are provided. A test using 800 images of Kelan River in
Xinjiang Province shows that this system can reduce 35%-50% time consumption in contrast of traditional methods, and
increases respond speed of UAV image processing rapidly.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
4
References
5
Citations
NaN
KQI