Description
Field monitoring of insect pests is fundamental in crop management to gain information about their presence and abundance in order to timely adopt proper actions to face the infestation and avoid economical losses. However, for some high-invasive pests of global importance like Halyomorpha halys, classical management procedures are ineffective due to high reproductive potential, high mobility and polyphagy (Maistrello et al., 2018). As an improvement for crop field management, spectral cameras mounted on Unmanned Aerial Vehicles (UAVs) and other IoT devices are becoming a promising innovative technology allowing fast, efficient, and real-time monitoring of insect infestations.
The present study has been developed in the frame of the HALY.ID project, which aims at implementing a prototype of a digital platform for monitoring the presence of brown marmorated stink bugs in crop fields. In this case, NIR hyperspectral imaging was used to overcome mimicry of H. halys and to identify the spectral wavebands more relevant for bug detection on different vegetal backgrounds.
The hyperspectral images were acquired in the 980-1660 nm range and then subjected to a masking procedure, performed by applying PCA, which permitted to effectively identify the pixel spectra of the bugs and those related to the different backgrounds. Based on these results, a library of reference spectra of H. halys bugs and of vegetal backgrounds was selected by Kennard-Stone algorithm and used for classification purposes using Soft Partial Least Squares-Discriminant Analysis (Soft PLS-DA) coupled with sparse based methods for spectral variable selection (Calvini et al., 2018). The classification models allowed to obtain satisfactory results in the detection of H. halys on different vegetal backgrounds. The selected spectral regions will be implemented in a multispectral imaging system, which is more suitable for automated on-field monitoring of the presence of H. halys.
Keywords: pest management, field monitoring, bug detection, spectral imaging, variable selection, multivariate classification
Acknowledgements: HALY.ID is part of ERA-NET Cofund ICT-AGRI-FOOD, with funding provided by national sources (Ministero delle politiche agricole e forestali, MIPAAF) and co-funding by the European Union’s Horizon 2020 research and innovation program, Grant Agreement number 862671.
REFERENCES
Calvini, R., Orlandi, G., Foca, G., Ulrici, A., 2018. Development of a classification algorithm for efficient handling of multiple classes in sorting systems based on hyperspectral imaging. Journal of Spectral Imaging. https://doi.org/10.1255/jsi.2018.a13
Maistrello, L., Dioli, P., Dutto, M., Volani, S., Pasquali, S., Gilioli, G., 2018. Tracking the spread of sneaking aliens by integrating crowdsourcing and spatial modeling: The Italian invasion of halyomorpha halys. BioScience 68, 979–989. https://doi.org/10.1093/biosci/biy112
