- Methodology
- Open Access
Multispectral imaging for presymptomatic analysis of light leaf spot in oilseed rape
- Received: 17 April 2018
- Accepted: 17 January 2019
- Published: 24 January 2019
Abstract
Background
The use of spectral imaging within the plant phenotyping and breeding community has been increasing due its utility as a non-invasive diagnostic tool. However, there is a lack of imaging systems targeted specifically at plant science duties, resulting in low precision for canopy-scale measurements. This study trials a prototype multispectral system designed specifically for plant studies and looks at its use as an early detection system for visually asymptomatic disease phases, in this case Pyrenopeziza brassicae in Brassica napus. The analysis takes advantage of machine learning in the form of feature selection and novelty detection to facilitate the classification. An initial study into recording the morphology of the samples is also included to allow for further improvement to the system performance.
Results
The proposed method was able to detect light leaf spot infection with 92% accuracy when imaging entire oilseed rape plants from above, 12 days after inoculation and 13 days before the appearance of visible symptoms. False colour mapping of spectral vegetation indices was used to quantify disease severity and its distribution within the plant canopy. In addition, the structure of the plant was recorded using photometric stereo, with the output influencing regions used for diagnosis. The shape of the plants was also recorded using photometric stereo, which allowed for reconstruction of the leaf angle and surface texture, although further work is needed to improve the fidelity due to uneven lighting distributions, to allow for reflectance compensation.
Conclusions
The ability of active multispectral imaging has been demonstrated along with the improvement in time taken to detect light leaf spot at a high accuracy. The importance of capturing structural information is outlined, with its effect on reflectance and thus classification illustrated. The system could be used in plant breeding to enhance the selection of resistant cultivars, with its early and quantitative capability.
Keywords
- Disease detection
- Light leaf spot
- Oilseed rape
- Multispectral
- Preprocessing
- Machine learning
- Support vector machine
- Novelty detection
- Orientation effects
- Photometric stereo
Background
The development of precision agriculture has brought about huge advances in monitoring technologies that allow for quantifiable and early detection of plant stress factors [1, 2]. Despite these technological advances and continuous improvement in plant varieties, yield improvements in many crops have plateaued in recent years [3]. This has been blamed in part on ineffective crop management, due to a lack of reliable tools for in-situ monitoring and intervention in increasingly varying conditions [4].
Plant disease is a leading contributor to global crop losses [5]. The selection of disease resistant crop varieties plays a central role in negating this diminution; typically achieved by visual scoring of symptom severity. Imaging techniques have considerable potential to improve this process, by enabling quantification of disease severity and development rate. This would facilitate breeding for crops with quantitative resistance; with individual traits and associated genes each contributing a small improvement to plant performance but, in combination, providing effective and sustainable disease resistance [6].
Plant diseases result in physiological and morphological alterations, with plant pigmentations caused by pathogen interactions [7]. In foliar fungal disease, infection is quantified by estimating the infected area on the leaf surface; a major limitation is the low accuracy due to the subjective nature of visual assessment [8]. Molecular methods have been exploited in plant pathology to improve the accuracy of diagnosis with serological assays, such as enzyme-linked immunosorbent assay (ELISA), used extensively for detection of pathogens [9] and nucleic acid-based methods, such as quantative polymerase chain reaction, based on using nucleotide primers with high affinity to target sequences. Although molecular methods are relatively low cost, they have disadvantages, as the sample preparation is destructive and labour intensive, and producing specific antibodies can be inefficient with the presence of inhibitors reducing the sensitivity of nucleic acid-based methods [10, 11]. As pathogens often do not spread uniformly inside plants, destructive molecular methods can be non-diagnostic, especially at the asymptomatic stage [10]. Therefore, new methods are required for precise non-invasive, non-destructive and continuous diagnosis.
The light leaf spot (LLS) pathogen (Pyrenopeziza brassicae) is an important disease of winter oilseed rape (OSR) (Brassica napus) [12–14]. This versatile crop has multiple uses, and is responsible for £0.7 billion of the UK agricultural market [15], with LLS responsible for losses of £400 million, between 2012 and 2014, with 95% OSR crops affected [16]. The onset of infection can be subtle, due to the hemibiotrophic nature of the pathogen. This makes in-situ analysis almost impossible during the initial infection stages [14, 17], relying on the emergence of late stage visual symptoms. Thus, novel approaches would benefit plant breeders seeking to develop new resistant varieties as well as farmers where early detection in the field opens the possibility of early intervention by sparse selective fungicide application, improving the efficacy of chemical applications.
Multispectral imaging (MSI) collect light reflected from the leaf surface. The spectrum of the reflected light is governed by both physical and biochemical interactions with the leaf, processes that will change during disease infection [18, 19]. Images are collected at different wavelengths in the optical range (350–1000 nm), where each pixel in the image represented by a vector, known as the spectral signature. Analysis typically uses spectral information to study plant properties and conditions through spectral vegetation indices (SVIs), used commonly in remote sensing to describe vegetation health and density [20]. Several SVIs have been developed to identify and detect plant disease [21, 22].
Using photometric stereo (PS) for analysing plant tissue has attracted limited attention in the literature. Recent work such as [23] has highlighted the usefulness of the approach for extracting high resolution 3D imagery of leaves in order to aid vein extraction. It has been shown that different wavelengths used for light sources affect the validity of the assumption that the surface exhibits Lambertian reflectance. For example in [24], reconstructions of human faces were shown to be more accurate under NIR than visible. This is thought to be due to the sub-surface scattering of NIR, which penetrates further into the skin, leading to a more diffuse reflectance. However, there is a compromise in using NIR for improving overall 3D reconstruction in that surface detail is lost. Whether the same issue affects plant tissue is not clear and will be the focus of further work, but for the purposes of this study visible light is used in order to extract as much higher resolution information as possible.
This study aims to develop a MSI system for disease diagnosis. This has been applied in a controlled environment for the analysis of disease progression and has the potential, with further development, to be used in the field [25]. This MSI system uses narrowband light sources and a broadband silicon detector, in contrast to typical MSI and HSI systems where a broadband illumination source is used with a narrow band (typically diffraction grating based) detector. This selected spectral resolution allows the technology to be applied at a fraction of the cost of traditional systems, whilst improving signal-to-noise ratio (SNR) due to targeted reflectance peaks. The use of machine vision algorithms allows for early symptom detection (before symptoms can be detected by eye) allowing better quantification of disease progression with the potential, in field production scenarios, to allow earlier and hence more effective intervention. The main contributions of this work lie in (1) introducing a refined MSI device, (2) detection of a hemibiotrophic infection before visible symptoms appear and (3) improvements in classification performance using tailored machine learning techniques.
Materials and methods
The aim of this study was to demonstrate the detection capabilities of MSI whilst investigating the effect of monitoring at canopy scape versus leaf scale. Detached leaf and entire plant assays were undertaken to provide information with samples at different angles as found within canopies versus the orthogonal orientation achieved using detached leaves and compare the time to diagnosis between the two conditions.
Host: oilseed rape
OSR (Brassica napus L.) has been selected as the host for these trials as it is an important crop and its broad leaf nature is useful to demonstrate the potential of canopy-scale imaging. Seeds (Department of Biology, University of York) were placed in Petri dishes on wet filter paper for three days until germination, when the seedlings were transferred to M3 (Levington) compost in \(80\times 80\times 90\) mm square pots. The plants were grown at \(16\,{^{\circ }}\)C with a photo-period of 12h light/12h dark at 75% humidity for two weeks prior to inoculation.
Pathogen: light leaf spot
Pathogen life-cycle evolution. Pyrenopeziza brassicae infection necrotic development on leaf tissue showing early phase senescence (top) and lesions from a late-stage sample (bottom) both showing a colour-coded stage of infection. Included for information is a \(\times\)40 microscopy evaluation from each sample
Trials
In order to investigate the pathogen-host interaction, alongside sensing limitations, two trials were used at different perspectives with plants and detached leaves.
Canopy assay
Scanning set-up. Side-view diagram of apparatus set-up for canopy and PS imaging (left) and detached leaf assay with MSI (centre) showing the major system components with three-dimensional view (right)
Detached leaf assays
Four OSR genotypes, Bristol(2), Charger(4), Cracker(9), Temple(7), were selected on the basis of their resistance, as per their given ratings (/10) [30]. The irradiance in the growth chamber was \(160\;\upmu \hbox {mol m}^{-2}\hbox { s}^{-1}\). At 25 DAI the inoculated leaves were removed with a sterile scalpel and placed on petri dishes within plastic trays with a small amount of distilled water. A small piece of tissue paper was wrapped around the leaf petiole and dipped into the water to keep the leaves moist. The study used 6 inoculated replicates and 6 mock-controls, from each of the four different genotypes. Data was collected at 26, 28, 31 and 34 DAI. Images were taken of leaf samples held in place using a wire mesh, see Fig. 2.
Imaging equipment
A multispectral imaging prototype, see Fig. 2, developed at the e-Agri Sensors Centre, University of Manchester was used to obtain the data presented in this study. Detailed information on the system operation can be found in Veys et al. [25]. The system can be used with multiple NIR sensitive detectors; the ones used for this trial were an Omnivision 5647 and a Sony IMX219; both included a Bayer filter, whose response artefacts were removed via calibration [25]. The device contained 36 narrowband sources from 365–960 nm, full width half maximum (FWHM) 10 nm, illuminated sequentially. This transferred considerably less energy to the sample, whilst having better SNR, than the broadband illumination used in passive MSI imaging; this is important as the non-invasive nature of MSI is not realised if significant heat is transferred to the samples. The illumination was provided by back projecting lensed LEDs into a barium sulphate and latex [31] solution coated dome, which allowed homogeneous lighting of the sample. The current minimum spatial resolution is between 0.1 and 0.2 mm/pix at an object distance of 300 mm. Dark-field images were subtracted from the multispectral cube during the acquisition. The raw data were acquired onto an on-board USB-3 storage device during the trial, before being transferred to a host computer for analysis. The data are then normalised using a calibration file, created by using a barium sulphate white tile as a maximum target to prevent saturation of the image sensor and remove slight variation due to the LED intensity and non-linear camera response [25].
Analysis software
The following sections include descriptions, with fundamental theory, on different processing and classification algorithms, which were written in MATLAB.
Background removal
Region of interest identification
Data extraction
Spectral signatures were automatically extracted, using random indexing within the identified regions of interest of each replicate. For the analysis 690 pixels were extracted for entire plants, 150 control and 540 inoculated, and 435 pixels were extracted for individual leaves, 222 control and 213 inoculated. Each of these samples was labelled whether from an inoculated or control plant to aid classifier training and allow calculation of error. The number of samples extracted was increased until there was minimal variation present within each replicate set. The standard deviation within the sub-sampled information was negligible for this number of samples, which can been seen in the standard deviation in the classification.
Spectral processing
These remove a large proportion of noise variation and help to mitigate the scattering artefacts of reflectance imaging. Without the pre-processing steps, then the classification algorithms will start to differentiate the variance due to unwanted artefacts such as distance from camera variations in leaf angle.
Data redundancy reduction
Conventional classification
The value of decision function \(f(x)\in {\pm 1}\) in which 1 denotes one class and \(-1\) represents the other class. The classification rate was represented as an average of 100 independent runs, showing the number of times a pixel was misclassified as inoculated.
One-class classification
3D reconstruction of plants
There are many other methods for reconstructing the surface from normals such as using the Discrete Cosine Transform to enforce integrability instead of the Fourier basis [43] or the M-estimator approach of [44], but the Frankot–Challappa method was chosen due to its efficiency and proven performance.
Results
Average classification rate of LLS in OSR (Charger) using spectral indices and selected wavelength at canopy and leaf scale
Input | Average classification rate % (std) | |
---|---|---|
Plant | Leaf | |
NDVI [45] | 36.6 (0.02) | 63.1 (0.02) |
PSRI [46] | 40.5 (0.04) | 61.9 (0.02) |
DWSI [47] | 49.0 (0.04) | 55.1 (0.02) |
CLS [22] | 52.2 (0.06) | 65.6 (0.03) |
CTR1 [34] | 52.4 (0.05) | 67.9 (0.03) |
ARI [48] | 56.8 (0.05) | 70.5 (0.02) |
PRI [49] | 59.5 (0.04) | 55.3 (0.02) |
LLSI [this study] | 59.5 (0.04) | 75.0 (0.02) |
MSI spectra | 60.4 (0.05) | 71.6 (0.02) |
FS spectra | 62.4 (0.05) | 75.3 (0.02) |
Multispectral backscattered reflectance
Infection Detection. Pathogen ingress detection demonstrated using the LLS for leaf and CTR1 for plant to outline disease severity, shown as a percentage, and distribution across a representative subset of the trial dates. The SVI colormap is normalised between 0 and 1 to display the infection intensity and distribution. The spatial scale is 1:25 with the 10 mm grid shown
Reflectance Spectra. Reflectance spectra of ROI, identified by LLSI for detached leaves (left) and CTR1 for entire plants (right) normalised to a control spectra at each time-point. The FS wavelengths are highlighted to show points of differentiation
Support vector machine classification
Classification rate improvement of using ND-SVM compared to conventional SVM for FS spectra of LLS in OSR entire plants (Charger)
DAI | Average classification rate % (std) | |
---|---|---|
SVM | ND-SVM | |
03 | 61.6 (0.04) | 82.6 (0.07) |
06 | 64.4 (0.04) | 83.4 (0.03) |
09 | 66.7 (0.04) | 88.3 (0.02) |
12 | 73.3 (0.03) | 92.2 (0.01) |
15 | 75.0 (0.03) | 92.3 (0.02) |
21 | 78.8 (0.04) | 93.5 (0.01) |
24 | 91.8 (0.02) | 94.5 (0.02) |
27 | 94.2 (0.03) | 94.7 (0.02) |
31 | 97.7 (0.03) | 96.7 (0.03) |
Classification rate comparison using conventional SVM for each cultivar with the resistance rating (/10) shown on fixed leaf data
DAI | Average classification rate % (std) | |||
---|---|---|---|---|
Bristol (2) | Charger (4) | Cracker (9) | Temple (7) | |
26 | 69.3 (0.03) | 72.9 (0.05) | 57.1 (0.06) | 73.7 (0.06) |
28 | 68.0 (0.04) | 75.2 (0.06) | 60.6 (0.06) | 69.3 (0.07) |
31 | 70.0 (0.04) | 77.8 (0.08) | 67.6 (0.09) | 70.2 (0.08) |
34 | 74.2 (0.06) | 78.5 (0.04) | 75.8 (0.11) | 77.0 (0.09) |
Classification rate comparison using ND-SVM for each cultivar with the resistance rating (/10) shown on fixed leaf data
DAI | Average classification rate % (std) | |||
---|---|---|---|---|
Bristol (2) | Charger (4) | Cracker (9) | Temple (7) | |
26 | 82.8 (0.04) | 50.8 (0.07) | 81.8 (0.04) | 82.9 (0.05) |
28 | 81.7 (0.05) | 61.8 (0.08) | 82.5 (0.05) | 74.0 (0.07) |
31 | 85.3 (0.07) | 77.2 (0.07) | 80.6 (0.06) | 81.2 (0.06) |
34 | 87.0 (0.10) | 80.7 (0.12) | 83.8 (0.08) | 87.0 (0.11) |
Novelty detection
Due to the localised nature of infection, the amount of data given to the SVM for infected tissue, despite ROI was limited, which constrained classification performance (see Table 1). Thus novelty detection (ND) was applied, using control plants for training and looking for instances that fall outside the accepted control variance. The FS wavelengths from the canopy dataset were classified using SVM and ND-SVM with the improvement shown in Table 2. The comparison for fixed leaves is shown by Tables 3 and 4.
It should be noted that whilst this method enhanced the rate of detection in this instance, it does not distinguish between disease and other stresses, not present in control plants, that may cause the reflectance to deviate beyond the devised limits. Hence, in practice, it could be used to detect areas of interest before a conventional SVM is applied to classify the type of stress; using a library of labelled responses. This is of particular interest due to the high classification rate from very early in the trial (09 DAI).
Although a larger population is required for phenotyping applications, an example of the difference in resistance of the four selected cultivars can be found in Table 4. Note that by tailoring the classifier to the correct type, the performance is improved significantly as the training sets become more representative with similar features and infection rates.
3D plant reconstruction
Orientation effects. Variation in reflectance measured at high (blue) [\(50\,{^{\circ }}\)] and low (red) [\(10\,{^{\circ }}\)] inclination normalisation (left) PS reconstruction of control oilseed rape plant, at 03 DAI with excessive curvature of leaves highlighted with arrows (right)
Discussion
The classification accuracy of the plant analysis (83-97%) was in-line with existing literature (61–99%) [20] however, the time from inoculation to 90+% diagnosis was a significant improvement (12 DAI), particularly due to the clandestine nature of the hemibiotrophic pathogen at early stages of its life-cycle. The active MSI technique allowed detection of disease regions allowing for localised classification of infected tissue areas. The method, with application of machine learning algorithms, allowed infection detection for entire plants at 92% accuracy 13 days earlier than visual inspection. The authors would urge caution with the early rates of classification for ND-SVM, caused by the difference in residue left from the control and pathogen inoculum. Although this is of interest in this study, it is a similar infection mechanism that could occur in-field or during high-throughput trials, which did not give any visible symptoms and thus outlines the requirement for spectral applications.
The dual-scale nature of the experiment allowed for a detailed investigation into the orientation effects that are so problematical in the HSI imaging of plant material [51–53]. The detached-leaf assay also had a number of drawbacks despite being the measurement of reference for flat samples. This difference in the time-scales between the two experimental set-ups, was chosen to reduce the development of chlorosis on detached leaf assay but also meant the infection took longer to manifest; resulting in visual symptoms seven days later for detatched leaves. The simulated view of the entire leaf on a flat plane gave an important insight into the pathogen mechanism on a leaf scale (see Fig. 3) and gave an example of what could be seen on the canopy, should orientation reflections be fully compensated. Since the leaf images were taken alongside the canopy, the significance of this effect on detection rates can be examined (see Fig. 5) with an average increase in classification rate of 15% for flat orthogonal samples.
The 3D capabilities are still in the testing phase as there are a number of limitations to the current design. The PS method manages to represent the surface features with high-fidelity but suffers on discontinuous surfaces; to negate this an OSAVI mask is applied to remove non-plant shapes from the scan, but in future, the inclusion of structured light will prevent this and allow for more macro surface orientation based morphology to be extracted. The current LED design means that a Gaussian distortion is introduced into the reconstruction, which can be seen in the model in Fig. 5; as the curvature of the leaf tip is exaggerated.
The study found that automated FS can play an important role in improving classification performance, when compared to SVIs or even full spectra with SVM. This is due to the limitation of specific wavelengths with mixed relevance to disease ingress and additional data-points obfuscating the classification respectively. A comparison demonstrating the improvement is shown in Table 1. Another important finding was that the improvement in prediction accuracy between the conventional (SVM) and one-class (ND-SVM) classifiers, illustrated in Table 2.
There are not sufficient replicates of the cultivars to be able to infer a clear relationship between the resistance and classification rates. However, since these rates have been defined on visual symptoms [30] then they would not be a good reference when compared to destructive molecular techniques. This system would need some modification before it could be utilised outside the lab environment as the performance would be affected by additional stresses (e.g. nutrient deficiency) and ambient lighting; thus the current design may prove of more value in controlled indoor environments.
Future work
There are a number of potential improvements to improve the data acquisition process, with a major focal point on fully incorporating the canopy orientation models into the spectral processing to remove the variation seen in Fig. 5. A faster global shutter system could replace the rolling shutter sensor; allowing a significant improvement in SNR. On-line visualisation of the multispectral cube and 3D information would inform the user of current limitations of imaging technique, which is often not found until the data is analysed. Finally, to improve the static nature of the current system, a common communication interface to a translational stage would allow integration in high-throughput applications. A revised system will be implemented in a field setting, incorporated into the spot-spraying equipment. This location has the added benefit of the shade created by the health and safety enclosures on all operational farm machinery, which in turn improves the SNR.
Conclusions
In this paper, a low-cost active MSI system has been developed. Different analysis techniques have been used with the primary goal of evaluating the application of this system to the study of plant pathogen interactions. The findings clearly indicate the ability to detect disease using spectral information, with the minimum detection level affected by leaf orientation. The paper also exploits machine learning methods to extend the diagnosis beyond user supervised techniques. The outcome of this will not only help to detect the onset of disease but will also help in breeding varieties in the future by extending current breeding capabilities, by allowing for better differentiation of resistant cultivars.
Declarations
Authors' contributions
CV designed the system and data processing techniques and was the major contributor in writing the manuscript. FC carried out the plant material inoculation, growth and imaging. JH engineered the printed circuit boards and provided manuscript revisions. GB and MH developed the 3D reconstruction technique. AA developed the feature selection and support vector classifications. FC, SR, CV and BG were involved in the trial design. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to acknowledge: Paul Shaw, University of Manchester, for his mechanical construction; Anthony Diggle, University of Manchester, for his circuit population abilities; Ian Bancroft, University of York, for providing the OSR seeds; Neil Havis, Scotland’s Rural College, for providing the LLS isolates; Jon West, Rothamsted Research, for providing valuable information regarding the pathogen inoculation.
Competing interests
The authors declare that they have filed a patent (GB1721451) surrounding the apparatus for determining this spectral and morphological information.
Availability of data and materials
The MSI and PS datasets for both trials undertaken are available in RAW format, compatible with all ENVI enabled software packages, from Mendeley Data (https://doi.org/10.17632/ydmtggnzbw.1).
Funding
The research was funded as part of EPSRC PhD studentships. FC would like to thank the Grantham Centre for Sustainable Futures for support. Further work at the University of Sheffield was supported by The Wolfson Foundation.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Authors’ Affiliations
References
- Mahlein AK, Oerke EC, Steiner U, Dehne HW. Recent advances in sensing plant diseases for precision crop protection. Eur J Plant Pathol. 2011;133(1):197–209. https://doi.org/10.1007/s10658-011-9878-z.View ArticleGoogle Scholar
- Zhang Q. Precision agriculture technology for crop farming, 1st edn. CRC Press, Washington; 2015. https://doi.org/10.1017/CBO9781107415324.004. arXiv:1011.1669v3.
- Grassini P, Eskridge KM, Cassman KG. Distinguishing between yield advances and yield plateaus in historical crop production trends. Nat Commun. 2013;4:1–11. https://doi.org/10.1038/ncomms3918.View ArticleGoogle Scholar
- Agriculture and Horticulture Development Board (AHDB): Oilseed rape guide. Technical Report January, HCGA 2014.Google Scholar
- Oerke E-C. Crop losses to pests. J Agric Sci. 2006;144(01):31. https://doi.org/10.1017/S0021859605005708.View ArticleGoogle Scholar
- St. Clair DA. Quantitative disease resistance and quantitative resistance loci in breeding. Ann Rev Phytopathol. 2010;48(1):247–68. https://doi.org/10.1146/annurev-phyto-080508-081904.View ArticleGoogle Scholar
- West JS, Bravo C, Oberti R, Moshou D, Ramon H, McCartney HA. Detection of fungal diseases optically and pathogen inoculum by air sampling. In: Oerke E-C, editor. Precision crop protection—the challenge and use of heterogeneity; 2010. pp. 135–149. Springer, Harpenden. https://doi.org/10.1007/978-90-481-9277-9-9.
- Niks RE, Parlevliet JE, Lindhout P, Bai Y. Breeding crops with resistance to diseases and pests. Wageningen: Wageningen Academic; 2011. p. 198.View ArticleGoogle Scholar
- Musgrave DR. Detection of an endophytic fungus of Lolium perenne using enzyme-linked immunosorbent assay (ELISA). N Z J Agric Res. 1984;27(2):283–8. https://doi.org/10.1080/00288233.1984.10430431.View ArticleGoogle Scholar
- Martinelli F, Scalenghe R, Davino S, Panno S, Scuderi G, Ruisi P, Villa P, Stroppiana D, Boschetti M, Goulart LR, Davis CE, Dandekar AM. Advanced methods of plant disease detection: a review. Agron Sustain Dev. 2015;35(1):1–25. https://doi.org/10.1007/s13593-014-0246-1.View ArticleGoogle Scholar
- McCartney HA, Foster SJ, Fraaije BA, Ward E. Molecular diagnostics for fungal plant pathogens; 2003. https://doi.org/10.1002/ps.575.
- Fitt BDL, Doughty KJ, Gilles T, Gladders P, Steed JM, Su H, Sutherland KG. Methods for assessment of light leaf spot (Pyrenopeziza brassicae) on winter oilseed rape (Brassica napus) in the UK. Ann Appl Biol. 1998;133(3):329–41. https://doi.org/10.1111/j.1744-7348.1998.tb05816.x.View ArticleGoogle Scholar
- Boys EF, Roques SE, Ashby AM, Evans N, Latunde-Dada AO, Thomas JE, West JS, Fitt BDL. Resistance to infection by stealth: Brassica napus (winter oilseed rape) and Pyrenopeziza brassicae (light leaf spot). Eur J Plant Pathol. 2007;118(4):307–21. https://doi.org/10.1007/s10658-007-9141-9.View ArticleGoogle Scholar
- Sharif B, Makowski D, Plauborg F, Olesen JE. Comparison of regression techniques to predict response of oilseed rape yield to variation in climatic conditions in Denmark. Eur J Agron. 2017;82:11–20. https://doi.org/10.1016/j.eja.2016.09.015.View ArticleGoogle Scholar
- DEFRA: British Food and Farming at a glance. Technical Report March, Department for Environment Food & Rural Affairs; 2016.Google Scholar
- Science Fera. CropMonitor: Disease Survey Highlights. Fera Science: Technical report; 2017.Google Scholar
- Ashby AM. A molecular view through the looking glass: the Pyrenopeziza Brassicae to Brassica Interaction. Adv Bot Res. 1997;24:32–70.Google Scholar
- Karolewski Z. Development of light leaf spot on Brassicas. Phytopathologia. 2010;55:13–20.Google Scholar
- Karolewski Z, Kosiada T. Sensitivity of Pyrenopeziza brassicae (Light Leaf Spot) populations to fungicides recommended for the fungus control (Sadowski 1989). Phytopathologia. 2010;56:59–65.Google Scholar
- Lowe A, Harrison N, French AP. Hyperspectral image analysis techniques for the detection and classification of the early onset of plant disease and stress; 2017. https://doi.org/10.1186/s13007-017-0233-z.
- Rumpf T, Mahlein AK, Steiner U, Oerke EC, Dehne HW, Plümer L. Early detection and classification of plant diseases with support vector machines based on hyperspectral reflectance. Comput Electron Agric. 2010;74(1):91–9. https://doi.org/10.1016/j.compag.2010.06.009.View ArticleGoogle Scholar
- Mahlein AK, Rumpf T, Welke P, Dehne HW, Plümer L, Steiner U, Oerke EC. Development of spectral indices for detecting and identifying plant diseases. Rem Sens Environ. 2013;128:21–30. https://doi.org/10.1016/j.rse.2012.09.019.View ArticleGoogle Scholar
- Zhang W, Hansen MF, Smith M, Smith L, Grieve B. Photometric stereo for three-dimensional leaf venation extraction. Comput Ind. 2018;98:56–67. https://doi.org/10.1016/j.compind.2018.02.006.View ArticlePubMedPubMed CentralGoogle Scholar
- Hansen MF, Atkinson GA, Smith LN, Smith ML. 3D face reconstructions from photometric stereo using near infrared and visible light. Comput Vis Image Underst. 2010;114(8):942–51. https://doi.org/10.1016/j.cviu.2010.03.001.View ArticleGoogle Scholar
- Veys C, Hibbert J, Davis P, Grieve B. An ultra-low-cost active multispectral crop diagnostics device. In: IEEE Sensors; 2017. pp. 1–3. IEEE, Glasgow. https://doi.org/10.1109/ICSENS.2017.8234211.
- Li D, Ashby AM, Johnstone K. Molecular evidence that the extracellular cutinase Pbc1 is required for pathogenicity of Pyrenopeziza brassicae on oilseed rape. Mol Plant Microbe Interact MPMI. 2003;16(6):545–52. https://doi.org/10.1094/MPMI.2003.16.6.545.View ArticlePubMedGoogle Scholar
- Rawlinson CJ, Sutton BC, Muthyalu G. Taxonomy and biology of Pyrenopeziza brassicae sp. nov. (Cylindrosporium concentricum), a pathogen of winter oilseed rape (Brassica napus ssp. oleifera). Trans Br Mycol Soc. 1978;71(3):425–39. https://doi.org/10.1016/S0007-1536(78)80070-9.View ArticleGoogle Scholar
- Davies KA, De Lorono I, Foster SJ, Li D, Johnstone K, Ashby AM. Evidence for a role of cutinase in pathogenicity of Pyrenopeziza brassicae on brassicas. Physiol Mol Plant Pathol. 2000;57(2):63–75. https://doi.org/10.1006/pmpp.2000.0282.View ArticleGoogle Scholar
- Karolewski Z, Evans N, Fitt BDL, Todd AD, Baierl A. Sporulation of Pyrenopeziza brassicae (light leaf spot) on oilseed rape (Brassica napus) leaves inoculated with ascospores or conidia at different temperatures and wetness durations. Plant Pathol. 2002;51(5):654–65. https://doi.org/10.1046/j.1365-3059.2002.00746.x.View ArticleGoogle Scholar
- Home Grown Cereals Authority (HGCA): Recommended Variety Lists. Technical report, HGCA; 2014.Google Scholar
- Knighton N, Bugbee B. A mixture of barium sulfate and white paint is a low-cost substitute reflectance standard for spectralon. Tech Instrum. 2005;11:4–6.Google Scholar
- Rondeaux G, Steven M, Baret F. Optimization of soil-adjusted vegetation indices. Rem Sens Environ. 1996;55(2):95–107. https://doi.org/10.1016/0034-4257(95)00186-7.View ArticleGoogle Scholar
- Otsu N. A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern 1979;9(1), 62–66. https://doi.org/10.1109/TSMC.1979.4310076. arXiv:1011.1669v3.
- Carter GA, Miller RL. Early detection of plant stress by digital imaging within narrow stress-sensitive wavebands 5 3. Rem Sens Environ. 1994;302:295–301.View ArticleGoogle Scholar
- Mahlein A-K, Steiner U, Hillnhütter C, Dehne H-W, Oerke E-C. Hyperspectral imaging for small-scale analysis of symptoms caused by different sugar beet diseases; 2012. https://doi.org/10.1186/1746-4811-8-3.
- Liu H, Yu L. Toward integrating feature selection algorithms for classification and clustering. IEEE Trans Knowl Data Eng. 2005;17(4):491–502. https://doi.org/10.1109/TKDE.2005.66.View ArticleGoogle Scholar
- Hall MA, Smith LA. Feature selection for machine learning: comparing a correlation-based filter approach to the wrapper. In: Proceedings of the twelfth international FLAIRS conference. AAAI Press; 1999.Google Scholar
- Cortes C, Vapnik V. Support-vector networks. Mach Learn 1995;20(3), 273–297. https://doi.org/10.1023/A:1022627411411. arXiv:1011.1669v3.
- Clifton L, Clifton DA, Zhang Y, Watkinson P, Tarassenko L, Yin H. Probabilistic novelty detection with support vector machines. IEEE Trans Reliab. 2014;63(2):455–67. https://doi.org/10.1109/TR.2014.2315911.View ArticleGoogle Scholar
- Hales IJ, Williamson DR, Hansen MF. Long-range concealed object detection through active covert illumination. In: SPIE Security + Defence, SPIE Publications, Toulouse 2015; vol. 9648, pp. 21–24. https://doi.org/10.1117/12.2190194.
- Frankot RT, Chellappa R. A method for enforcing integrability in shape from shading algorithms. IEEE Trans Pattern Anal Mach Intell. 1988;10(4):439–51. https://doi.org/10.1109/34.3909.View ArticleGoogle Scholar
- Brooks M, Horn B. Shape and source from shading. In: Horn BKP, Brooks MJ, editors. Shape from shading. Cambridge: MIT Press; 1989. p. 53–68.Google Scholar
- Georghiades AS, Belhumeur PN, Kriegman DJ. From few to many: illumination cone models for face recognition under variable lighting and pose. IEEE Trans Pattern Anal Mach Intell. 2001;23(6):643–60.View ArticleGoogle Scholar
- Agrawal A, Raskar R, Chellappa R. What is the range of surface reconstructions from a gradient field? In: European conference on computer vision; 2006, pp. 578–591. https://doi.org/10.1007/11744023-45.
- Rouse JW, Haas RH, Schell JA, Deering DW. Monitoring vegetation systems in the Great Okains with ERTS. Third Earth Resour Technol Satell Symp. 1973;1:325–33.Google Scholar
- Merzlyak MN, Gitelson AA, Chivkunova OB, Rakitin VYU. Non-destructive optical detection of pigment changes during leaf senescence and fruit ripening. Physiol Plant. 1999;106(1):135–41. https://doi.org/10.1034/j.1399-3054.1999.106119.x.View ArticleGoogle Scholar
- Apan A, Held A, Phinn S, Markley J. Detecting sugarcane ‘orange rust’ disease using EO-1 Hyperion hyperspectral imagery. Int J Rem Sens. 2004;25(2):489–98. https://doi.org/10.1080/01431160310001618031.View ArticleGoogle Scholar
- Gitelson A, Merzlyak M, Chivkunova OB. Optical properties and nondestructive estimation of anthocyanin content in plant leaves. Photochem Photobiol. 2001;74(1):38–45. https://doi.org/10.1562/0031-8655(2001).View ArticlePubMedGoogle Scholar
- Gamon JA, Serrano L, Surfus JS. The photochemical reflectance index: an optical indicator of photosynthetic radiation use efficiency across species, functional types, and nutrient levels. Oecologia. 1997;112(4):492–501. https://doi.org/10.1007/s004420050337.View ArticlePubMedGoogle Scholar
- Amigo JM, Babamoradi H, Elcoroaristizabal S. Hyperspectral image analysis: a tutorial. Anal Chim Acta. 2015;896:34–51. https://doi.org/10.1016/j.aca.2015.09.030.View ArticlePubMedGoogle Scholar
- Paulus S, Behmann J, Mahlein AK, Plümer L, Kuhlmann H. Low-cost 3D systems: suitable tools for plant phenotyping. Sensors (Switzerland). 2014;14(2):3001–18. https://doi.org/10.3390/s140203001.View ArticleGoogle Scholar
- Liew OW, Chong PCJ, Li B, Asundi AK. Signature optical cues: emerging technologies for monitoring plant health. Sensors. 2008;8:3205–39. https://doi.org/10.3390/s8053205.View ArticlePubMedGoogle Scholar
- Alsuwaidi A, Veys C, Hussey M, Grieve B, Yin H. Hyperspectral selection based algorithm for plant classification. In: IEEE instrumentation and measurement society; 2016, p. 16.Google Scholar