Detection of the granary weevil based on X-ray images of damaged wheat kernels

Grain in storage is exposed to a number of adverse factors, including extensive damage to grain kernels caused by infestations of the granary weevil Sitophilus granarius. This pest causes a major decline in grain quality leading to a substantial drop in the value of the stored material, thus contributing to large financial losses. It is therefore essential to ensure that this pest is identified promptly and accurately if present in the stored grain. The purpose of this study was to define the visual representative features found in digital X-ray images of wheat kernels that bear traces of inner kernel damage caused by the granary weevil. Such features are required to build training sets, which are crucial for the development of digital neural classifiers. Subsequently, a set of identifying neural models was produced and verified, after which an optimal topology was selected. The optimal artificial neural network (ANN) was a three-layer perceptron with the following structure: 8:11-6-1:1. The proposed model identified 100% of the infested kernels correctly, and 98.4% of the healthy ones. The analysis of the sensitivity of the generated neural model demonstrated the significance of the following three graphical parameters determining the quality of damaged kernel identification: cultivar, Feret coefficient (WF) and the area (P) of the kernel. •The study was based on neural network modeling methods, including image analysis.•Artificial neural networks are a powerful tool in the identification of granary weevil infestation.•The training set was generated on the basis of previously acquired X-ray images.•The best classification ability was achieved by MLP Topology.