| |
return to Farmers
Independent Weekly
June
26, 2003
By Jitendra Paliwal, Department of Biosystems Engineering |
U
of M researchers test machine vision systems for grain grading
About 90 years ago, when the grading system of grains, pulses
and oilseeds was established in Canada, it was assumed that
grains that looked better were better in quality. Appearance
was a very subjective trait and there were no devices available
to measure appearance. Therefore, the visual system of grading
grain was devised and adopted, based on five grading factors:
test weight, varietal purity, soundness, vitreousness and
maximum limit of foreign material. The latter four factors
are determined visually by trained personnel. To further differentiate
among the different classes of wheat grown in Canada, parameters
such as Kernel Visual Distinguishability (KVD) were also developed.
This
system has worked very effectively to establish Canada as
a major supplier of quality grain in the international grain
market. However, the manual system of grading grains has drawbacks.
Although grain inspectors go through rigorous training, grading
large samples of grain is a very tedious job and decisions
can be influenced by various factors such as experience and
expertise of the personnel, working conditions, fatigue, etc.
A faster objective system of grading grain is desirable and,
with the advancement of technology, we now have the tools
to automate the grading process.
The
technology of machine vision, which has arisen from a union
between camera and computer, has the capability to identify
and classify different objects. In a machine vision system
(MVS) a video camera acts as an eye and the computer does
the work of the brain. Machine vision offers many advantages
over the conventional grading systems. It is compatible with
other automated on-line processing tasks, can work round the
clock, can take dimensional measurements more accurately and
consistently than human beings, and can give an objective
measure of variables such as color, projected area, and shape
which an inspector could only assess subjectively. Since the
inspection is done without contact, it is hygienic and there
is less damage to the fragile biological products being inspected.
Because
the shape, size and color of cereal grains are dependent on
such variables as their area of production, levels of maturity,
growth and harvest conditions, using MVS to identify and classify
cereal grains has been a challenge. A team of grain storage
experts at the Department of Biosystems Engineering, University
of Manitoba, has been working to develop technologies that
would overcome these challenges. Researchers have used various
mathematical concepts such as Fourier transforms and run-length
matrices to characterize kernels of different types of cereal
grains. The software to classify and grade grains is available
now and work is being done to automate and mechanize the process
of sample collection and image acquisition. Very soon we will
have a system that will be capable of collecting grain samples
automatically from railcars, storage bins or conveyor belts
and transport the samples to an MVS location for analysis
and grading. The whole process will require very little human
intervention and hence will assist the grain inspectors in
their job of grading grain.
We
are also working on designing a grain cleaner that will use
machine vision to identify contaminants present in a sample
prior to and after passing through the cleaner. The cleaner
parameters can then be adjusted to optimize its performance.
In
a nutshell, the technology of MVS will assist grain inspectors
in grading grain and the resulting objective grading system
will help Canada in maintaining and expanding its role as
a dominant player in the international grain market.
|
|
