The one place for your designs
To enable design management, you'll need to meet the requirements. If you need help, reach out to our support team for assistance.
weird results of otbcli_DimensionalityReduction (pca)
Description
Resulting PC image has striping. Observed in several different images.
Result very different from that obtained with R (RSToolbox::rasterPCA())
(in both cases, RGB composites of PC1,PC2,PC3)
Also, eigenvectors are different:
otb:
BertMICA20190531v2PCA.csv
R:
eigenvR.csv
(note: eigenvectors in R by columns)
In both cases, input image has no data values set to 0:
https://drive.google.com/open?id=1JBeC2khACkbJc1Hit-ZVDia6X7IS2yV8
Note: PCA for large images is impractical in R because it takes too long a time.
Steps to reproduce
otbcli_DimensionalityReduction.bat -in BertMICA20190531v2.tif -outBertMICA20190531v2PCA.tif uint8 -nbcomp 5 -normalize false -outmatrix BertMICA20190531v2PCA.csv -ram 128
Configuration information
otb 6.6.1 (same results on windows and linux)