X
PrevPrev Go to previous topic
NextNext Go to next topic
Last Post 29 Nov 2021 11:00 AM by  MariM
ISODATA Classification
 1 Replies
Sort:
You are not authorized to post a reply.
Author Messages

Federico Carotenuto



New Member


Posts:3
New Member


--
26 Nov 2021 04:01 AM
    Dear all,

    I am trying to do some unsupervised classification on hyperspectral images using the ISODATA tool. However, I find some unclarity between the GUI options and the ENVI help.

    ISODATA requires to specify thresholds about standard deviation of the data ("Maximum class Stdv") and the help states that this standard deviation should be expressed in "DN" (which I guess means "Digital Number"). But what does it means exactly? I am working on scaled reflectance images, how should I convert them to DN?

    Also the default standard deviation threshold is 1, which seems odd even if working with DNs. Does this mean that ISODATA performs some kind of normalization on the data?

    Thanks for the clarification!

    MariM



    Veteran Member


    Posts:2396
    Veteran Member


    --
    29 Nov 2021 11:00 AM
    DN is used generically here to represent 'digital number' in whatever units your data is in. If the data has not been calibrated to radiance or reflectance, then enter the distance, SD, etc as DN in the original units. If the data has been converted to reflectance (0-1.0), then enter the distance or other parameters DN as floating point values (for example, a min distance of .0.2 and max of 0.4).
    You are not authorized to post a reply.