Wavelet Transform Applied to Coffee Entomology

• Autor(a) principal: Lemos Escola, Joao Paulo
• Data de Publicação: 2021
• Outros Autores: Da Silva, Ivan Nunes, Guido, Rodrigo Capobianco [UNESP], Fonseca, Everthon Silva
• Tipo de documento: Artigo de conferência
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1109/SPSympo51155.2020.9593404; http://hdl.handle.net/11449/234042

In this work, the design and development of a computational algorithm to assist in the management of insect pests in coffee plantations are presented, particularly for detecting the presence of cicadas. Acoustic signals, previously captured, are submitted to the proposed system which reads the raw data, converts them to the wavelet domain and groups them together based on the Bark Scale. Then, Paraconsistent Characteristics Analysis, appearing as a technique recently presented in the scientific literature and which had not yet been used for this purpose, serves as a basis for selecting the best filter banks so that they can be later delivered to a Support Vector Machine (SVM), responsible for the final step of signal identification. The accuracy of 100% was achieved in most of the 3600 tests performed, proving the viability of the implemented strategy, which has become minimally complex due to the optimization provided by the paraconsistent methodology. Finally, a prototype in the scope of Internet of Things is described to serve as a possibility of implantation in the field.