The tuna processing industry currently uses individual fish weight to estimate fish
size. The correlation between fish weight and other physical properties are important
because they could also be used in both size grading methods and in heat transfer
modeling. Table 2 shows the regression equations and correlation coefficient (R2) obtained
between fish weight and other physical properties determined in this study. All but one
of the relationships studied exhibit large correlation coefficients (R2 = 0.74–0.97), includ-
ing volume, projected area (side and top view), length, width (measured at the thickest
part), and perimeter (measured at the thickest part). These correlations are illustrated in
Figs. 4a–4h.
Only the relationship between fish weight and thickness at the thickest part of the
fish was relatively poor (R2 = 0.49). Strong correlation between fish weight and fish pro-
jected area (top view and side view) were found (R2 = 0.92 and 0.94, respectively), which
indicates that an image processing grader could be an attractive alternative grading method.
In such a grading method, the side view projected area image would be preferable to the top
view, because the fish could be measured while traveling on their sides on a conveyor belt.
Figure 4h shows the linear relationship between fish volume and surface area (R2 = 0.84).
This information could be used in the fish wrapping process as well. When the volume of
a fish is known, the surface area can be calculated from the relationship equation, and the
required wrapping area can be estimated.