03-19-2019 02:18 PM
I'm looking for a example spatial model where an Independent Component Analyses (ICA) is apllied on geospatial data. Did somebody already worked with the ICA operator in the spatial modeler.
Currently I'm using "ICA" operator in combination of the "Anomely Detection" operator, and wondering whether this is the proper method. Some addiotional documentation in the operator help would be usefull.
03-28-2019 08:13 AM
The ICA operator computes a matrix from a n-band image which can be used with the LinearCombination operator to produce a second n-band (or less) image in which the first band contains most of the information in the image, the second band less and the last band would contain the least information (i.e. noise).
The Anomaly Detection operator examines an n-band image and computes an output single band output image in the which higher values are more anomalous.
So is your thinking that you would like to feed the AnomalyDetection from the output of the ICA (scenario-1) or are you thinking about feeding the output of Anomaly Detection into the ICA (scenario-2).
In scenario-1 you would use the matrix from ICA to drive a Linear Comination of the original input and get a new n-band output which could then be fed into Anomaly Detection.
Scanario-2 will not work since the output of Anomaly Detection is a single band and ICA require an input of more than one band.
Can you describe what you are trying to achieve with the combination of these operators?
04-02-2019 06:37 AM