|
|
Detekce napadení ječmene listovými chorobami pomocí infračerveného termálního zobrazování, zobazovací fluorescence chlorofylu, UV-stínění fluorescence chlorofylu a hyperspekrálního zobrazování
Holzmannová, Kateřina
The aim of the thesis was, in the theoretical part, to summarize basic information about barley, its use and the leaf diseases that occur most often on barley, and also to summarize the basics of imaging methods potentially useful for the detection of diseases on plant leaves, namely chlorophyll fluorescence imaging, chlorophyll fluorescence UV shielding , hyperspectral imaging and infrared thermal imaging. The aim of the work was to evaluate the potential of individual sensor methods from the point of view of their ability to detect the resistance of spring barley genotypes to powdery mildew based on the experiment carried out with the inoculation of spring barley with grass powdery mildew (Blumeria graminis f. sp. hordei). In the practical part, selected sensor methods are compared within the experiment with 6 lines of spring barley (line of the Pallas variety) differing only in the presence of different genes for resistance to powdery mildew. These lines were chosen to represent a wide range of reaction types in response to infection with avirulent powdery mildew pathotypes, as reaction type can be reflected in the response measured using these indirect methods. Based on the correlation analysis performed for individual parameters measured using tested indirect sensor methods with the values of the light-saturated CO2 assimilation rate, it was possible to state that within each group of methods there is a parameter showing a high ability to predict the plant's response and thus resistance to powdery mildew. It is a chlorophyll index, based on the principle of measuring the infrared and red transmittance of the leaf, the difference between the leaf temperature and the air temperature determined using a thermal camera, the fluorescence decrease ratio - vitality index (Rfd), and finally also the ZM spectral index. In addition to the temperature difference, the values of the Pearson correlation coefficient reach the level of 0.74, which means that it is a highly evident relationship, and the adjusted R2 values are above 0.55, which means that the reliability of the prediction is higher than 55%.
|
|
|
Detekce houbových chorob ječmene pomocí zobrazovací fluorescence a termálního zobrazování
Holzmannová, Kateřina
The main goal of this work was to perform a literature review of imaging methods for the detection of spring barley diseases, their advantages and disadvantages and to conduct an experiment on spring barley samples with different levels of fungal disease infection to compare disease detection using chlorophyll fluorescence imaging, UV fluorescence imaging and hyperspectral imaging in VNIR and NIR spectral ranges. The originally planned measurements on more diseases were performed due to governmental travel restrictions related to the coronavirus pandemic only for Pyrenophora teres barley infection. The achieved results show that all three methods have the potential for use in the field of early and non-invasive detection of P. teres infection, either for the purpose of aerial monitoring or in the phenotyping of resistance to P. teres. The best correlation to leaf infection was achieved with the parameter actual quantum yield of photosystem II (ΦPSII). This parameter shows a high linearity of relationship to P. teres infection, which gives the assumption of reliable detection in the whole infection range. With regard to the need for measurements from the proximity and also due to possible effect of changes in the intensity of solar radiation, this is rather a parameter applicable for plant phenotyping than for the monitoring of large areas. Very good results in the detection of P. teres infestation were also obtained using the UV fluorescence screening index. UV fluorescence screening also shows a very close relationship to the level of P. teres infection, but of a non-linear nature, where the UV fluorescence shielding index increases sharply with under a low level of infection and is less affected at higher level of infection. This means a higher sensitivity of the method to a small level of infection and may therefore have an advantage especially in the early detection of infection. The UV screening index also have considerable potential for the analysis of the disease spatial distribution at the leaf level, where UV shielding increases in areas around the infestation site but decreases significantly at the infestation site itself. In contrast, hyperspectral reflectance represents considerable potential for monitoring of large areas, because it can be evaluated not only from proximity but also from a drones, aircrafts or satellites. The highest correlation and linearity of relationship between P. teres infection was achieved for the index integrating the wider reflectance range 650-725 nm ANMB650-725. However, good results of estimating the level of infection were also achieved with simple vegetation indices such as NDVI and EVI. For the early detection of P. teres infection, the ARI index based on the anthocyanin absorption band, showing a sharp change in low infection level, seems to be the most suitable. This index also shows very good results of evaluating the spatial distribution of P.teres infection on the leaf, with ARI rising at the site of infection. The best results of the evaluating the spatial distribution of infection was demonstrated by the simple ratio of reflectances in the red and green band (RGR), which, however, shows less close relationship to the total P. teres on the whole leaf. Overall, it can be stated that all three methods have the potential to be used for monitoring of P. teres infection and for phenotyping of resistance to P. teres. For the selection of method, rather the key technical requirements e.g. if the total leaf infection is measured or the distribution of the leaf infection is evaluated, from what distance the measurement can be performed and what is the range of measured areas, and also whether it is necessary to detect low level of infection with high sensitivity at the onset of infection or it is more important to ensure a reliable estimate of the infection in the whole range.
|
guest ::
