Classification of Processing Damage in Sugar Beet ( Beta vulgaris ) Seeds by Multispectral Image Analysis.

Abstract

The pericarp of monogerm sugar beet seed is rubbed off during processing in order to produce uniformly sized seeds ready for pelleting. This process can lead to mechanical damage, which may cause quality deterioration of the processed seeds. Identification of the mechanical damage and classification of the severity of the injury is important and currently time consuming, as visual inspections by trained analysts are used. This study aimed to find alternative seed quality assessment methods by evaluating a machine vision technique for the classification of five damage types in monogerm sugar beet seeds. Multispectral imaging (MSI) was employed using the VideometerLab3 instrument and instrument software. Statistical analysis of MSI-derived data produced a model, which had an average of 82% accuracy in classification of 200 seeds in the five damage classes. The first class contained seeds with the potential to produce good seedlings and the model was designed to put more limitations on seeds to be classified in this group. The classification accuracy of class one to five was 59, 100, 77, 77 and 89%, respectively. Based on the results we conclude that MSI-based classification of mechanical damage in sugar beet seeds is a potential tool for future seed quality assessment.

Plant phenotyping relevance

サトウダイコン種子の機械的損傷という植物器官の状態を、多波長画像と統計モデルで分類する方法を評価しており、種子品質評価のための画像ベース表現型計測が中心である。

Code and data availability

The article describes multispectral imaging of sugar beet seeds and an nCDA/SAS classification model, but contains no public dataset, image, or code deposit. Seed samples were provided by MariboHilleshög (proprietary), analysis used commercial VideometerLab software and SAS, and no data or code availability statement,,

No evidence-backed public reproduction asset is currently recorded.