Image Analysis of Pellet Size for a Control System in Industrial Feed Production
Citation: Ljungqvist MG, Nielsen ME, Ersbll BK, Frosch S (
Image Analysis of Pellet Size for a Control System in Industrial Feed Production
Martin Georg Ljungqvist 0
Michael Engelbrecht Nielsen 0
Bjarne Kjaer Ersbll 0
Stina Frosch 0
Guy J-P. Schumann, University of Bristol, United Kingdom
0 1 Department of Informatics and Mathematical Modelling, Technical University of Denmark (DTU), Kongens Lyngby, Denmark, 2 Division of Industrial Food Research, National Food Institute, Technical University of Denmark (DTU) , Kongens Lyngby , Denmark
When producing aquaculture fish feed pellets, the size of the output product is of immense importance. As the production method cannot produce pellets of constant and uniform size using constant machine settings, there is a demand for size control. Fish fed with feed pellets of improper size are prone to not grow as expected, which is undesirable to the aquaculture industry. In this paper an image analysis method is proposed for automatic size-monitoring of pellets. This is called granulometry and the method used here is based on the mathematical morphological opening operation. In the proposed method, no image object segmentation is needed. The results show that it is possible to extract a general size distribution from an image of piled disordered pellets representing both length and diameter of the pellets in combination as an area.
-
In the aquaculture industry it is of outmost importance that the
fish get feed of proper size. The feed is usually in pellet form, where
the pellets contain the nutrients that the fish need to grow and stay
healthy. The size of the pellets is adapted to the size of the fish so that
the fish can grow as expected. It has been shown that growth rate of
fish is closely related to the pellet size of feed [13]. Therefore, when
producing feed pellets for aquaculture there is a need to control the
size of the output product, and this is a challenging task.
An extruder machine is commonly used for fish feed production.
The feed material is extruded through a die plate with holes of a
certain diameter which determines the diameter of the pellets. On
the other side of the disk is a set of rotating knives that cut the
material into shorter cylinder-shaped pellets. The length of the
pellets is affected both by the velocity of the knives and the pressure
inside the machine.
When the extruder machine has been running for some time, the
holes in the die plate get clogged with raw material. The time-frame
for this to happen depends on the composition of the raw material
and the pellet size produced. This clogging of holes restrains further
material either completely or partly from flowing through the
holes and therefore affects the output of the machine.
Additionally, the pressure inside the extruder rises during
operation, inducing a rise in the velocity of the feed as it passes
through the holes, resulting in a drift of the pellet size. Moreover,
the temperature of the machine increases, and this might affect its
physical properties, both velocity and pressure. Changes in
pressure and temperature result in the problem that the pellet
size changes over time during a batch production.
Today, size monitoring is done by manual inspection in order to
adjust the settings or restart the machine. This is both
labourdemanding and relies on experienced assessors. An automatic
vision system for on-line quality control would be of great benefit
to the industry, both for process control and product optimisation.
If automatic size measurement could indicate when the pellet size
is outside the defined range, this information could be used to
adjust machine settings such as the knife speed, screw speed, filling
rate or other means of controlling the pellet expansion process,
thereby controlling the pellet size and ensuring uniformity.
Measuring the size distribution of small particles is often referred
to as granulometry or sometimes as particle size distribution analysis
[4]. The proposed method is based on image analysis using
mathematical morphology and in particular using the so-called
morphological openings used for size distribution analysis. This
technique was proposed by Matheron (1975) [5], a vast amount of
work on granulometry on binary images was also done by Serra
(1982) [6] and the technique has been further developed for
greyscale images [7,8].
Morphological openings are widely used for granulometry in
image analysis and have been used for many applications [913].
These all use an image segmentation method before performing
the morphological opening.
Another approach to granulometry is the use of frequency
transform analysis, as can be seen in Zadoro z_ny et al. (2002) [14],
where a specific segmentation method is not needed. The same
paper also used the technique of scale-space [1517], which can
likewise be seen in Clemmensen et al. (2009) [18]. A similar
approach without segmentation can be seen in Jagersand (1995)
[19].
Measuring fish feed (...truncated)