The application of neural network data mining algorithm into mixed pixel classification in geographic information system environment
With the rapid growth of satellite technology and the increasing of spatial resolution, hyperspectral imaging sensor is frequently used for research and development as well as in some semi-operational scenarios. The hyperspectral image also offers unique applications such as terrain delimitations,...
Saved in:
| Main Author: | |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | en |
| Published: |
2007
|
| Subjects: | |
| Online Access: | http://eprints.utem.edu.my/id/eprint/15156/1/The%20application%20of%20neural%20network%20data%20mining%20algorithm%20into%20mixed%20pixel%20classification%20in%20geographic%20information%20system%20environment214.pdf http://eprints.utem.edu.my/id/eprint/15156/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | With the rapid growth of satellite technology and the
increasing of spatial resolution, hyperspectral imaging sensor is frequently used for research and development as well as in some semi-operational scenarios. The hyperspectral image also offers unique applications such as terrain delimitations, object detection, material identification, and atmospheric characterization. However, hyperspectral image systems produce large data sets that are not easily interpretable by visual analysis and therefore require automated processing algorithm. The challenging of pattern recognition associated with hyperspectral images is very complex processing due to
the presence of considerable number of mixed pixels. This
, paper discusses the development of data mining and pattern
recognition algorithm to handle the complexity of
hyperspectral remote sensing images in Geographical
Information Systems environment. Region growing
segmentation and radial basis function algorithms are
considered a powerful tool to minimize the mixed pixel
classification error. |
|---|