Rice (Oryza sativa L.) is one of the major commercial
grains worldwide, the economic value of which strongly
depends on the proportion of unbroken kernels. In the rice
processing industry, the quality of final product depends
on several sensory properties. Among the later, visual
properties are more important because they can significantly
affect the choice and preferences of consumers.
Commonly, there are two major visual indices for determining
the quality of rice kernels in the processing industry
namely, degree of milling (DOM) and percentage of
broken kernels (PBK). Generally, in rice mills, due to
unavailability of continuous on-line measurement methods,
quality grade of the product is monitored visually by
experienced operators at 1–2 h intervals [1]. This means
that the operator, based on his experience and proficiency
with the processing machinery, assesses the quality of the
product by mere visual inspection of the machine output
and making the required adjustments. In this regard,
development of automated systems which can mimic the
expert operators’ decision-making process would be quite
beneficial for quality grading of the product. Nevertheless,
such an automated quality grading of rice kernels is not an
easy task.
Soft computing is an innovative method for development
of intelligent systems which has attracted increasing
interest by the scientific community during the past few
decades. Generally, in order to solve real world computation
problems, a combination of computational techniques
is preferred to the exclusive use of single methods. One
such combinative method is neuro-fuzzy system [2]. Since
its inception, the theory of fuzzy sets has advanced in a
variety of ways in many disciplines. Applications of this
theory can be found, for example, in artificial intelligence,
computer science, medicine, control engineering, decision