PLS_Toolbox Documentation: imgsimca	< imgselct	isimcapr >

imgsimca

Purpose

SIMCA Classification for Multivariate Images.

Synopsis

model = imgsimca(img,data)

newmod = imgsimca(data,model)

Description

This function allows the user to build SIMCA class models by outlining areas on a reference image img that correspond to areas on a multivariate image data. This image must be the same size (pixels by pixels) as the multivariate image data but can have a different depth (possibly due to having preprocessed the original data with IMGPCA).

The output is a structure array model which may be applied to new images with ISIMCAPR or IMGSIMCA. The fields of model are:

               imgname:   name of original image variable,

               mwaname:   name of original multivariate image data,

                     name:   model type, always 'IMGSIMCA',

                     date:   date stamp,

                     time:   time stamp,

                     size:   size of the image data,

             simcamod:   simca model parameters (see SIMCA),

             in_class:   array which gives pixels within 95% Q limits for each class,

   closest_class:   array which gives closest class for each pixel,

                   rqmat:   reduced Q values for all classes for all pixels, and

               rtsqmat:   reduced T² values for all classes for all pixels.

Examples

To construct an image simca (IMGSIMCA) model

      model = imgsimca(img,data);

To make class predictions on a new multivariate image data use the model developed in IMGSIMCA and then call IMGSIMCA as

      newmod = imgsimca(data,model);

See Also

contrastmod, imgpca, imgselct, isimcapr