Correction of the Virtual Dimensionality (VD) algorithm. This fixes the bug #1841 (closed).
In the virtual dimensionality algorithm, according to [1] (this is the reference cited in the code), the variance of the noise should be estimated as sigma = 2/N * (lambdaK^2 + lambdaR^2)
instead of sigma = 2/N * (lambdaK + lambdaR + lambdaK*lambdaR)
. lambdaK
and lambdaR
being the sorted eigenvalues of the covariance and the correlation matrices, respectively.
[1] C.-I. Chang and Q. Du, "Estimation of number of spectrally distinct signal sources in hyperspectral imagery", IEEE Transactions on Geoscience and Remote Sensing, vol. 43, no. 3, mar 2004.
The baselines for virtual dimensionality (class and application) has been modified
The copyright owner is CNES and has signed the ORFEO ToolBox Contributor License Agreement.
Check before merging: