DocumentCode
1854989
Title
Multi-channel texture classification applied to feature extraction in forestry
Author
Körber, Christian ; Möller, Dietmar P F ; Kätsch, Christoph
Author_Institution
Hamburg Univ.
fYear
2005
fDate
22-25 May 2005
Lastpage
6
Abstract
This paper suggests an alternative image processing method that may lead to determining the species, age, health, and size of trees in mixed forest stands from aerial images. The approach is based on k-means clustering of per-pixel signatures which are derived from several band passes of the FFT coefficients of two small moving windows. A meaningful segmentation of the clusters is obtained by interactively classifying the clusters to classes
Keywords
fast Fourier transforms; feature extraction; forestry; image classification; image segmentation; image texture; pattern clustering; FFT coefficients; aerial images; cluster segmentation; feature extraction; forestry; image processing; image segmentation; k-means clustering; multichannel texture classification; per-pixel signatures; Agriculture; Costs; Data mining; Digital images; Feature extraction; Forestry; Image processing; Image segmentation; Image texture analysis; Remote monitoring;
fLanguage
English
Publisher
ieee
Conference_Titel
Electro Information Technology, 2005 IEEE International Conference on
Conference_Location
Lincoln, NE
Print_ISBN
0-7803-9232-9
Type
conf
DOI
10.1109/EIT.2005.1626988
Filename
1626988
Link To Document