Limits...
Hyperspectral and thermal imaging of oilseed rape (Brassica napus) response to fungal species of the genus Alternaria.

Baranowski P, Jedryczka M, Mazurek W, Babula-Skowronska D, Siedliska A, Kaczmarek J - PLoS ONE (2015)

Bottom Line: The measurements of disease severity for chosen dates after inoculation were compared to temperature distributions on infected leaves and to averaged reflectance characteristics.Statistical analysis revealed that leaf temperature distributions on particular days after inoculation and respective spectral characteristics, especially in the SWIR range (1000-2500 nm), significantly differed for the leaves inoculated with A. dauci from the other species of Alternaria as well as from leaves of non-treated plants.The second-derivative transformation of the spectral data together with back-propagation neural networks (BNNs) appeared to be the best combination for classification of days after inoculation (prediction accuracy 90.5%) and Alternaria species (prediction accuracy 80.5%).

View Article: PubMed Central - PubMed

Affiliation: Institute of Agrophysics, Polish Academy of Sciences, Lublin, Poland.

ABSTRACT
In this paper, thermal (8-13 µm) and hyperspectral imaging in visible and near infrared (VNIR) and short wavelength infrared (SWIR) ranges were used to elaborate a method of early detection of biotic stresses caused by fungal species belonging to the genus Alternaria that were host (Alternaria alternata, Alternaria brassicae, and Alternaria brassicicola) and non-host (Alternaria dauci) pathogens to oilseed rape (Brassica napus L.). The measurements of disease severity for chosen dates after inoculation were compared to temperature distributions on infected leaves and to averaged reflectance characteristics. Statistical analysis revealed that leaf temperature distributions on particular days after inoculation and respective spectral characteristics, especially in the SWIR range (1000-2500 nm), significantly differed for the leaves inoculated with A. dauci from the other species of Alternaria as well as from leaves of non-treated plants. The significant differences in leaf temperature of the studied Alternaria species were observed in various stages of infection development. The classification experiments were performed on the hyperspectral data of the leaf surfaces to distinguish days after inoculation and Alternaria species. The second-derivative transformation of the spectral data together with back-propagation neural networks (BNNs) appeared to be the best combination for classification of days after inoculation (prediction accuracy 90.5%) and Alternaria species (prediction accuracy 80.5%).

No MeSH data available.


Related in: MedlinePlus

False-colour compositions of visible and near infrared (VNIR) (503.5 nm, 660.1 nm, and 970.5 nm) and short wavelength infrared (SWIR) (1200.3 nm, 1759.2 nm, and 2202.3 nm) hyperspectral bands of an oilseed rape leaf inoculated with Alternaria alternata. The lower plot shows spectral characteristics within selected regions of the image representing the infected area, uninfected area, and entire leaf with standard deviation (std).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4380467&req=5

pone.0122913.g006: False-colour compositions of visible and near infrared (VNIR) (503.5 nm, 660.1 nm, and 970.5 nm) and short wavelength infrared (SWIR) (1200.3 nm, 1759.2 nm, and 2202.3 nm) hyperspectral bands of an oilseed rape leaf inoculated with Alternaria alternata. The lower plot shows spectral characteristics within selected regions of the image representing the infected area, uninfected area, and entire leaf with standard deviation (std).

Mentions: Spectral characteristics of the leaves indicated high differentiation of their reflectance response in the areas with the symptoms of Alternaria infection and in the uninfected areas. In Fig 6, exemplary false-colour images of oilseed rape leaves 3 d after inoculation in VNIR (left) and SWIR (right) ranges are presented together with reflectance spectra in the selected areas. These three regions correspond to: 1, the part of the leaf with visible symptoms of infection; 2, the part of the leaf without any symptoms of infection (uninfected area); and 3, the area of the entire leaf. Additionally, the standard deviation lines for uninfected and infected regions are included. The reflectance in the infected area was higher than in the uninfected area in the entire VNIR-SWIR range, and non-overlapping standard deviation lines indicate good separation for the areas of the leaf with and without symptoms of the infection. In the visible range of the spectrum, the absolute mean difference values of the reflectance between areas with and without disease symptoms were highest from 545–700-nm wavelengths (which include the chlorophyll absorption red-edge sub-region). The best separation between the uninfected and infected areas is observed in the SWIR range and also in the water-absorption bands (1470 and 1900 nm).


Hyperspectral and thermal imaging of oilseed rape (Brassica napus) response to fungal species of the genus Alternaria.

Baranowski P, Jedryczka M, Mazurek W, Babula-Skowronska D, Siedliska A, Kaczmarek J - PLoS ONE (2015)

False-colour compositions of visible and near infrared (VNIR) (503.5 nm, 660.1 nm, and 970.5 nm) and short wavelength infrared (SWIR) (1200.3 nm, 1759.2 nm, and 2202.3 nm) hyperspectral bands of an oilseed rape leaf inoculated with Alternaria alternata. The lower plot shows spectral characteristics within selected regions of the image representing the infected area, uninfected area, and entire leaf with standard deviation (std).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4380467&req=5

pone.0122913.g006: False-colour compositions of visible and near infrared (VNIR) (503.5 nm, 660.1 nm, and 970.5 nm) and short wavelength infrared (SWIR) (1200.3 nm, 1759.2 nm, and 2202.3 nm) hyperspectral bands of an oilseed rape leaf inoculated with Alternaria alternata. The lower plot shows spectral characteristics within selected regions of the image representing the infected area, uninfected area, and entire leaf with standard deviation (std).
Mentions: Spectral characteristics of the leaves indicated high differentiation of their reflectance response in the areas with the symptoms of Alternaria infection and in the uninfected areas. In Fig 6, exemplary false-colour images of oilseed rape leaves 3 d after inoculation in VNIR (left) and SWIR (right) ranges are presented together with reflectance spectra in the selected areas. These three regions correspond to: 1, the part of the leaf with visible symptoms of infection; 2, the part of the leaf without any symptoms of infection (uninfected area); and 3, the area of the entire leaf. Additionally, the standard deviation lines for uninfected and infected regions are included. The reflectance in the infected area was higher than in the uninfected area in the entire VNIR-SWIR range, and non-overlapping standard deviation lines indicate good separation for the areas of the leaf with and without symptoms of the infection. In the visible range of the spectrum, the absolute mean difference values of the reflectance between areas with and without disease symptoms were highest from 545–700-nm wavelengths (which include the chlorophyll absorption red-edge sub-region). The best separation between the uninfected and infected areas is observed in the SWIR range and also in the water-absorption bands (1470 and 1900 nm).

Bottom Line: The measurements of disease severity for chosen dates after inoculation were compared to temperature distributions on infected leaves and to averaged reflectance characteristics.Statistical analysis revealed that leaf temperature distributions on particular days after inoculation and respective spectral characteristics, especially in the SWIR range (1000-2500 nm), significantly differed for the leaves inoculated with A. dauci from the other species of Alternaria as well as from leaves of non-treated plants.The second-derivative transformation of the spectral data together with back-propagation neural networks (BNNs) appeared to be the best combination for classification of days after inoculation (prediction accuracy 90.5%) and Alternaria species (prediction accuracy 80.5%).

View Article: PubMed Central - PubMed

Affiliation: Institute of Agrophysics, Polish Academy of Sciences, Lublin, Poland.

ABSTRACT
In this paper, thermal (8-13 µm) and hyperspectral imaging in visible and near infrared (VNIR) and short wavelength infrared (SWIR) ranges were used to elaborate a method of early detection of biotic stresses caused by fungal species belonging to the genus Alternaria that were host (Alternaria alternata, Alternaria brassicae, and Alternaria brassicicola) and non-host (Alternaria dauci) pathogens to oilseed rape (Brassica napus L.). The measurements of disease severity for chosen dates after inoculation were compared to temperature distributions on infected leaves and to averaged reflectance characteristics. Statistical analysis revealed that leaf temperature distributions on particular days after inoculation and respective spectral characteristics, especially in the SWIR range (1000-2500 nm), significantly differed for the leaves inoculated with A. dauci from the other species of Alternaria as well as from leaves of non-treated plants. The significant differences in leaf temperature of the studied Alternaria species were observed in various stages of infection development. The classification experiments were performed on the hyperspectral data of the leaf surfaces to distinguish days after inoculation and Alternaria species. The second-derivative transformation of the spectral data together with back-propagation neural networks (BNNs) appeared to be the best combination for classification of days after inoculation (prediction accuracy 90.5%) and Alternaria species (prediction accuracy 80.5%).

No MeSH data available.


Related in: MedlinePlus