abstract

• One of the techniques for plant diseases diagnosis is the analysis of the spectral signatures in the vegetation. Among spectral analysis, there are methods for feature extraction to find out an early detection considering a significant number of samples of leaves in different growing stages considering healthy leaves, leaves in germination stage of the fungus, the first symptoms and diseased leaves identified in a previous work. In this study, with multiple comparisons and statistical tests, the discrete wavelet and Fourier transforms are employed for the feature extraction of the spectral data. A total of 13 features based on wavelet and 17 features on Fourier transform have significant differences between powdery mildew levels. The decomposition of the spectral signatures by wavelet-based and frequency domain coefficients as features shows a promising base by the classification for plant disease detection in cucurbits plants. © 2021 IEEE.

authors

• Morales Saldaña Jorge Alberto
• Palacios Hernández Elvia Ruth

publication date

• 2021-01-01

published in

• 2021 23rd IEEE International Autumn Meeting on Power, Electronics and Computing, ROPEC 2021  Proceedings

keywords

• Diagnosis; Discrete wavelet transforms; Extraction; Fourier transforms; Frequency domain analysis; Fungi; Spectrum analysis; Wavelet decomposition; Discrete wavelets; Features extraction; Germination stage; Growing stages; Multiple comparisons tests; Multiple statistical tests; Number of samples; Plant disease diagnosis; Powdery mildew; Spectral signature; Feature extraction

Digital Object Identifier (DOI)

• 10.1109/ROPEC53248.2021.9668078

PubMed ID

start page

end page

volume

issue