Table 1.
Field pea calibration and validation sets.
Fig 1.
Model development flow chart.
Fig 2.
Image pre-processing and processing flow chart.
Table 2.
Seed characteristics extracted through image processing.
Table 3.
Linear discriminant analysis models and parameters.
Table 4.
Classification Rates of Models.
Fig 3.
Colour factor variations between field pea market grades.
Four representative samples from each market grade illustrate the variation in relative colour intensity factors. The violet, blue, green, orange, red and NIR colour factors for each sample are represented respectively by the violet diamonds, blue squares, green triangles, yellow squares, red squares and pink circles. These are the basis for predicting market grades through Model 1.
Fig 4.
(a) and (b) All samples that were correctly classified (red dots) fell along the one to one correlation line (green), i.e. the closest market grade mean was the correct market grade mean for that sample. All samples which did not lie on the green line (one to one correlation) were incorrectly classified (blue triangles). Plots (c) and (d) gave the same scatter plots as (a) and (b) but highlighted which samples were non-defective (red dots) and which were defective (blue triangles).