Abstract
To recognize species-specific symptoms of plant diseases automatically, a near surface imaging at a sub-millimeter scale is necessary. Current investigations relate to an inspection of the upper visible crop parts using camera imaging from above the canopy. The paper presents primarily results using a sensor system, which is operating inside the canopy to monitor the vertical health status of winter wheat. This in-canopy sensor is also inspecting lower, invisible crop parts. With this technology a spatial and temporal precise crop protection—a future spraying of only the infected field parts would be possible. A commercially available camera with an NIR filter was attached to a vertical rod. To protect the camera while moving through the crop, it was installed inside a tube. This tube guaranteed a constant distance to the crop tissue so that it did not cover the camera lens. The mobile device was flexibly mounted on the back, three-point linkage of a tractor. In this first use example, the sensor was operated inside the crop canopy to inspect the lower leaves and directly at the canopy surface to inspect the upper leaves and the ears. In the images of the leaves inside the canopy, symptoms of tan spot (anamorph Drechslera tritici-repentis (Died.) Shoemaker) were clearly visible at flowering. When the camera was operated at the canopy level, black ears (caused by black point/kernel smudge/black head molds) were visible at milk ripeness.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
According to the rules of good agricultural practice, farmers periodically inspect the health status of their fields throughout the growing season. Because manual scouting is time-consuming, the farmer often looks only at a few locations usually at the border of his fields. The major part of the field keep uninspected. If especially diseases are just present and the weather conditions are favorable for their development, than fungicide spraying may occur too late. Therefore, entire fields need to be monitored by mobile sensors.
Under European conditions, pesticide spraying in cereals is often done between shooting and flowering. A multitude of foot rot, foliar and ear diseases beside of mainly insect pests my reach economic importance to legitimate their chemical control. Often, it is difficult to distinguish between the different pathogens to avoid failures in pesticide usage. As example, various fungal diseases and insects may cause similar symptoms at the foot of cereal culms (Dammer 1987, 1990). In recent years, high-resolution camera-based identification of the causation of crop damage has been increasingly investigated. For this identification approach, cameras are attached on unmanned aerial vehicles (UAVs, Wiesner-Hanks et al. 2019) or ground vehicles (Ruckelshausen and Busemeyer 2015). In addition, a combination of UAVs and groundbased imaging for species-specific disease symptom recognition has been conducted (Bohnenkamp et al. 2019; Dammer et al. 2021). Sophisticated image analysis methods, e.g., using machine-learning algorithms (Behmann et al. 2015; Schirrmann et al. 2021), are applied progressively to identify symptoms within the images automatically. Regarding to economic and ecological aspects, for a demand related precise crop protection a spraying of uninfected areas makes no sense. Running a GPS additionally to sensors, a localization of diseased field areas would be possible. An intensive real-time detection of crop diseases and the combination with an application step contributes to precision agriculture in crop protection (Kuska and Mahlein 2018).
When farmers inspect their cereal fields, they usually open the canopy and look at the lower, older leaves, which are longer exposed to possible diseases. Primary infection usually occurs on leaves near the soil (Wiese 1987). In case of the early infections, diseases spread from older/lower leaves to younger/higher leaf levels; thus knowing the disease status below the crop canopy provides the farmer a time advantage. He can evaluate the inoculum for future infections. If lower leaves show disease symptoms, then there is an imminent risk for infections of healthy plant tissue. Image-based monitoring methods from above (Satellites, UAVs, and ground vehicles) can only provide an inspection of the health status of the upper visible leaves, not of those inside the canopy. Therefore, in this short communication, the concept of a novel, mobile in-canopy imaging system for detecting fungal diseases in cereals and a first use-example are presented, and a future use of this technology is discussed.
Materials and methods
Concept and assembling of the vertical sensor
A first vertical sensor, based on photodiodes, was developed to detect the dieback of leaves caused by stripe (yellow) rust (Puccinia striiformis Westend. f. sp. tritici) and leaf (brown) rust (Puccinia triticina Erikss.) in winter wheat below the crop surface (Dammer et al. 2021). Because of the mode of action (spectrometric reflectance measures) of this photodiode-based, in-canopy sensor, single symptoms and their spatial pattern could not be recognized. Therefore, discrimination of the origin (e.g. nutrient/water deficiency, viral/fungal diseases, insect infestation, and natural senescence/ripeness) of the leaf damage was not possible. In addition, to spectral characteristics, the camera imaging delivered shape parameters and could identify a cause of the symptoms. A monochrome approach offers also an option to obtain a gray scale image (for example: 8 bit, 0 to 555 Gy scale values), which is easy to process further, and the near infrared (NIR) wavelength shows clear differences between the reflection of fungal damaged and healthy plant material (Anonym 2001).
A commercially available camera (Nikon ONE J3) was modified by removing the IR/UV-blocking filter to allow the infrared light to reach the sensor of the camera. A lens of 10 mm focal length was used, and the camera was equipped with an NIR long pass filter (maximum wavelength: 830 nm). The camera was attached to a vertical rod (Fig. 1 left). To protect the camera from touching the cereal plants, the camera was installed within a tube (Fig. 1 right). Additionally, this tube guarantees a distance of the camera lens to the plant tissue of approximately 0.10 m. Otherwise, the tissue would be too close to the lens and would often cover it.
First use example
In 2021, a first test of the mobile vertical camera system was conducted in winter wheat. In autumn 2020, at the experimental station of the Leibniz Institute for Agricultural Engineering and Bioeconomy (ATB), the winter wheat variety Matrix® (Deutsche Saatveredelung–DSV) was seeded (210 kernels per m2, 10 cm distance between rows) in a 90 m × 42 m field (WGS84: 12.955° E, 52.467° N). This field is characterized by diluvial soils (sandy loam) and a relatively low annual precipitation (400–600 mm). All agricultural measures (fertilization, weed control, and growth regulator application) were performed according to the principles of good agricultural practice.
The test run was conducted twice along the 90 m side of the field, on June 23, 2021 at flowering and on July 05, 2021 at milk ripeness. The sensor was operated in the field approximately 1.30 m from the border. To obtain images from the first (F-1) and second (F-2) leaf below the flag leaf (F), that were not visible from above, the camera was approximately 0.5 m from the ground. A sensor system for evaluation of the crop health should be all-purpose. Beside foliar and ear diseases also insect pests may occur, especially on the upper leaves—for example the rose-grain aphid (Metopolophium dirhodum Walker). Although they can be seen also from above, other insects may be not. The cereal leaf beetle (Oulema melanopus (L.)) deposits it eggs on the leaf underside and the larvae of the first stage hatch there. Therefore, additionally, at the second run on July 05, 2021, the camera was lifted to a height of approximately 1.25 m from the ground to obtain images from the flag leaves and ears. The camera was set in video mode (Full HD 1920 × 1080/60i). A 60i video recording led effectively to 30 frames per second. Along the operation track of the camera, visual assessments of the winter wheat plants for disease symptoms on the leaves and ears were performed.
Results
In-canopy sensing
In visual inspections along the sensor track on the first measuring date of June 23 at flowering, only some symptoms of tan spot caused by the anamorph Drechslera tritici-repentis (Died.) Shoemaker were manually visible on the first and second leaf below the flag leaf. In the NIR images, single leaves were clearly recognizable. As example, a NIR image (Fig. 2) is shown. A single disease symptom at the apex of a first leaf (F-1) is visible. Because of the different reflection intensity in the NIR wavelength, a spatial structure of the symptom could be recognized. A clear dark (A) spot (low reflection) was located in the middle of the chlorotic surrounding (B) tissue (high reflection). A dark (C) strip (low reflection) was bordering all.
Approximately two weeks later, on July 5 at milk ripeness, only the flag leaves were still green; all lower leaves (only visible when the canopy was opened by hand) were dead and deformed (Fig. 3 top). Because of the low intensity of D. tritici-repentis infections, this was caused primary by increasing natural senescence. In the NIR-image (Fig. 3 bottom), those dead first and second leaves (F-1/F-2) were darker (low reflectance) than the intact green flag leaves (F). They were brighter (high reflectance).
At-canopy sensing
During the second test run on July 5 at milk ripeness, black points/black head molds/sooty molds (kernel smudge) occurred, which is usually caused by saprophytic fungal genera such as Alternaria, Cladosporium, Colletotrichum. The visual inspection of the crop along the sensor track showed that some ears were completely infected, and some ears had black points at the glumes. As example, Fig. 4 shows an NIR image. Infected glumes and ears were clearly visible as they were dark (low reflection) in comparison to the brighter (high reflection) healthy ears.
Discussion
The advantage of the presented in-canopy sensor is the generation of high-resolution images below the crop canopy. Within the presented two NIR images (Figs. 2 and 4), the symptoms of two diseases tan spot and black point/black head molds/sooty molds/kernel smudge were recognized as dark objects. They less reflected the NIR wavelength compared to the surrounding healthy crop tissue. As visual inspected along the sensor track, in case of D. tritici-repentis mostly a dark spot was surrounded by bright dead tissue. The dead tissue was dark encircled at the boarder to the healthy tissue. This is a characteristic feature for the unique visual assignment of the symptoms of D. tritici-repentis. In case of the ears, the visual inspection of the crop along the sensor track showed that some ears were completely infected and single glumes often showed black points.
It should be further considered, that during the final fungicide application at flowering, it is most important to protect the green leaf area of the upper three leave levels and the ear. If only the flag leaf (F) is intact at this time (as it is the case in the NIR-image in Fig. 3), the spray liquid needs to not reach lower ones, than the spray volume could be reduced. Regarding this spraying strategy, the sensor system has high potential for online-spraying with variable rates. Regarding to digital image analysis, the gray-scale values of the NIR-image pixels lay between 0 and 555 (8 bit) with the zero value representing white (Haberaecker 1989). The pixels of non-green plant tissue have higher gray-scale values (low reflection) than those of green plant tissue (high reflection). In a simple image analysis, this could be used to discriminate non-green tissue from green tissue by setting a threshold gray-scale value. In the second step in this type of analysis, morphological image analysis methods (Haberaecker 1989) or machine learning approaches (mentioned in the introduction chapter of this paper) could be applied to visualize the shape of the symptoms so that could be assigned to specific disease species. Providing additional details regarding the numerous image-analyzing methods is beyond the scope of this short communication.
Compared to disease monitoring from above the canopy, disease monitoring inside the canopy provides knowledge of disease species, which are present at older leaves, and provides the farmer a time advantage in terms of disease control. If the disease locations in the field are also known, than the farmer is able to spray site-specifically. In future, this could contribute to a precise crop protection in time and space. Diseases, insect pests and weeds can occur simultaneously, than it would be useful to spray various pesticide products in one operation cycle. This could be carried out by fast direct injection field sprayers, which are under development (Krebs et al. 2015; Pohl et al. 2019). In a future sensor design, several cameras (focused on the ground, stems, leaves, and ears) can be installed at the vertical rod to monitor a magnitude of pathogens.
A health status evaluation of the lower, older leaves conducted using the presented in-canopy sensor technology provides information about actual infection potential, which is important for the epidemiological spread of diseases. To contribute to the forecasting of disease outbreaks, this information could be included in disease forecast models or Decision Support Systems (DSSs).
If head blight/scab (Fusarium spp.) or Alternaria spp. infest not only the glumes, but also the kernels in the ear, than the harvested grain can contain Deoxynivalenol (Vomitoxin, DON) or Alternaria toxins (Anonym 2006). Conducting the site-specific health status of the ears could be a requirement for a precise target-oriented harvesting technology of grain in the future. Even combine harvesters with a real-time grain flow separation were under investigation (Risius and Korte 2010).
References
Anonym (2001) Precision Farming im Pflanzenschutz: am Beispiel Unkrautbekämpfung. KTBL-Schrift 402, Darmstadt
Anonym (2006) Grain fungal diseases & mycotoxin reference. United States Department of Agriculture (USDA) Grain inspection, packers & stockyards administration. https://tgmsystem.com/wp-content/uploads/2014/11/mycobook.pdf. Accessed 28 June 2022
Behmann J, Mahlein A-K, Rumpf T, Römer C, Plümer L (2015) A review of advanced machine learning methods for the detection of biotic stress in precision crop protection. Prec Agric 16:239–260. https://doi.org/10.1007/s11119-014-9372-7
Bohnenkamp D, Behmann J, Mahlein A-K (2019) In-field detection of yellow rust in wheat on the ground canopy and UAV scale. Rem Sens 11:2495. https://doi.org/10.3390/rs11212495
Dammer K-H (1987) Verwechslungsmöglichkeiten des Schadbildes der Halmbruchkrankheit (Pseudocercosporella herpotrichoides (Fron) Deighton) mit Symptomen verschiedener Schadinsekten des Getreides. Arch Phytopathol Pflanzenschutz 23:513–515
Dammer K-H (1990) Das Auftreten von Mykosen an der Halmbasis von Winterroggen. Nachr Pflanzenschutz DDR 44:149–153
Dammer K-H, Schirrmann M, Garz A (2021) Primarily tests of a optoelectronic in-canopy sensor for evaluation of vertical disease infection in cereals. Pest Manag Sci 78:143–149. https://doi.org/10.1002/ps.6623
Haberaecker P (1989) Digitale Bildverarbeitung. 3rd edn. Carl Hanser Verlag, München
Krebs M, Rautmann D, Nordmeyer H, Wegener JK (2015) Development of a direct injection system without time lag for application of plant protection products. Landtechnik 70:238–252. https://doi.org/10.15150/lt.2015.3114
Kuska MT, Mahlein A-K (2018) Aiming at decision making in plant disease protection and phenotyping by the use of optical sensors. Eur J Plant Pathol 152:987–992. https://doi.org/10.1007/s10658-018-1464-1
Pohl JP, Rautmann D, Nordmeyer H, Hörsten VD (2019) Direkteinspeisung im Präzisionspflanzenschutz-Teilflächenspezifische Applikation von Pflanzenschutzmitteln. Gesunde Pflanzen 71:51–55. https://doi.org/10.1007/s10343-019-00452-y
Risius H, Korte H (2010) Process analysis for grain flow segregation on a combine harvester. Landtechnik 65:34–37. https://doi.org/10.15150/lt.2010.452
Ruckelshausen A, Busemeyer L (2015) Toward digital and image-based phenotyping. In: Kumar J, Pratap A, Kumar S (eds) Phenomics in crop plants: trends, options and limitations. Springer, New Delhi, pp 41–60
Schirrmann M, Landwehr N, Giebel A, Garz A, Dammer K-H (2021) Early detection of stripe rust in winter wheat using deep residual neural networks. Front Plant Sci 12:469689. https://doi.org/10.3389/fpls.2021.469689
Wiese MV (1987) Compendium of wheat diseases. 2nd edn. The American Phytopathological Society, St Paul
Wiesner-Hanks T, Wu H, Stewart E, DeChant C, Kaczmar N, Lipson H, Gore MA, Nelson RJ (2019) Millimeter-level plant disease detection from aerial photographs via deep learning and crowdsourced data. Front Plant Sci 10:1550. https://doi.org/10.3389/fpls.2019.01550
Acknowledgements
Thanks go to the colleagues of the Department Engineering for Crop Production André Hamdorf and Uwe Frank for construction and measuring work.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Dammer, KH. Proof of concept study: a novel mobile in-canopy imaging system for detecting symptoms of fungal diseases in cereals. J Plant Dis Prot 129, 769–773 (2022). https://doi.org/10.1007/s41348-022-00638-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s41348-022-00638-z