index exists which is desirable by producers, increase in
DOM is accompanied with a somewhat irregular trend in
the resulting quality class.
3.4. FIS evaluation
In order to evaluate the performance of the developed
FIS classifier, intensity and dimensional features of 100
milled rice samples of each class assessed by human
experts were computed using the image processing algorithm
and employed as input variables for the FIS. Totally,
25 quality levels were prepared by combination of DOM
and PBK MFs. The performance of the FIS classifier was
evaluated by forming a confusion matrix (CM) and computing
the statistical parameters of sensitivity (Se), specificity
(Sp) and classification accuracy (Ac) using the
following equations [33,34]: