SwePub - search: id:"swepub:oai:slubar.slu.se:3...

Enumeration	Reference	Cover	Find
1.	Nordkvist, Karin, et al. (author) Classification of EBONE General Habitat Categories in a Swedish forest area by using LiDAR in combination with SPOT satellite data 2010 Reports (other academic/artistic)abstract We have investigated to which degree a combination of optical satellite data and LiDAR data can improve classification accuracy of the General Habitat Categories (GHC) used by the FP7 project European Biodiversity Observation Network (EBONE), compared to using satellite data alone. The study was carried out in Remningstorp, a forest dominated area in southern Sweden. Remote sensing data used in the study were a SPOT 5 image from August 2009 and a laser scanning (26 points/m2) from September 2008. Ground truth samples were collected by interpretation of color infrared digital air photos from September 2009. Maximum likelihood and Random Forests classifications were made with satellite data and with a combination of satellite and LiDAR data. The classification scheme consisted of six forest classes, arable land and pasture land. The use of LiDAR data improved over-all accuracy with 6% for maximum likelihood classification and 7% for Random Forests. The highest over-all accuracy was obtained with Random Forests, but on the expense of the smaller classes.

Nordkvist, Karin, et al. (author)
Classification of EBONE General Habitat Categories in a Swedish forest area by using LiDAR in combination with SPOT satellite data
2010
Reports (other academic/artistic)abstract
- We have investigated to which degree a combination of optical satellite data and LiDAR data can improve classification accuracy of the General Habitat Categories (GHC) used by the FP7 project European Biodiversity Observation Network (EBONE), compared to using satellite data alone. The study was carried out in Remningstorp, a forest dominated area in southern Sweden. Remote sensing data used in the study were a SPOT 5 image from August 2009 and a laser scanning (26 points/m2) from September 2008. Ground truth samples were collected by interpretation of color infrared digital air photos from September 2009. Maximum likelihood and Random Forests classifications were made with satellite data and with a combination of satellite and LiDAR data. The classification scheme consisted of six forest classes, arable land and pasture land. The use of LiDAR data improved over-all accuracy with 6% for maximum likelihood classification and 7% for Random Forests. The highest over-all accuracy was obtained with Random Forests, but on the expense of the smaller classes.

Träfflista för sökning "id:"swepub:oai:slubar.slu.se:32699" "