Method for identifying soilborne pathogens for target crops in a field plot - Patents.com

Abstract

The present invention relates to a method for identifying a soil-borne pathogen for a target crop on an agricultural land plot, the method comprising the steps of: acquiring a first digital image of the agricultural land plot during a first harvest cycle, where a target crop is grown on the agricultural land plot during the first harvest cycle; acquiring a reference digital image of the agricultural land plot during a reference harvest cycle, where a reference crop is grown on the agricultural land plot during the reference harvest cycle, the reference crop being different from the target crop; calculating a first vegetation index associated with a first pixel in the first digital image; determining a first signed distance between the first pixel and pixels surrounding the first pixel based on the first vegetation index, and detecting a first anomaly for the first pixel if the first signed distance is less than a defined threshold; defining a reference anomaly for the reference pixel of the reference digital image; and identifying a soil-borne pathogen for the first pixel if the first anomaly does not match the reference anomaly.

Description

[Technical field]

[0001] The present invention relates to a method for identifying soil borne pathogens for target crops in an agricultural plot, and in particular to a remote sensing method for mapping and managing nematode pressure in soybean. More particularly, the present invention relates to a method for remote sensing assessment of nematode damage in soybean fields, and more generally to a method for crop health monitoring in the field. [Background technology]

[0002] Identifying and monitoring soil-borne pathogens such as diseases or pests is crucial to ensure high productivity in the fields. These pathogens can generally survive in the soil for several years and can rapidly multiply in the presence of the host crop. Nematodes are plant parasitic soil-borne pathogens that can infect more than 200 different plant species from 23 families. They live in the soil and feed on plant roots. Soybean crops are host to about 100 nematode species, among which the soybean cyst nematode (Heterodera glycines), root-knot nematode (Pratylenchus brachyurus), root-knot nematode (Meloidogyne incognita / javanica), reniform nematode (Rotylenchus reniformis), and spiral nematode (Helicotylenchus spp.) are the most common. The life cycle of the nematodes varies slightly depending on the species; however, they all feed on the root system.

[0003] Nematodes occur within plots of fields and have low mobility. Crop damage can occur when nematodes penetrate and feed on soybean roots, reducing the plant's ability to access water and nutrients, resulting in reduced productivity. Infection symptoms above ground are not unique to nematode infections and can be confused with nutrient deficiencies, stress from drought, phytotoxicity, or other pests and diseases. This is why methods to assess nematode-associated stress are limited. Traditional methods rely on human scouting with expert visual assessment of the typical spotty shape of nematode stress. Such assessments rely on the trained assessor's eye, which can be biased, inaccurate, non-repeatable, and do not scale to large geographical areas. Finally, the presence of nematodes can also be confirmed by local soil sampling and laboratory procedures to identify different nematode species and count the number of individuals. Again, such techniques do not scale well to large geographical areas, are costly, and may not capture the local spatial variability of nematode populations. Countermeasures to reduce the impact of nematodes include agricultural practices such as crop rotation, use of resistant varieties, and use of nematicides. In particular, crop rotation involves alternating soybeans with other non-host crops or plants that are less susceptible to nematodes, such as, but not limited to, corn, to reduce nematode populations in the soil.

[0004] Therefore, there is a need for improved methods for mapping soil-borne pathogens such as nematode stress in crops that are reliable and effective, especially over large geographical areas, which would allow for better soil-borne pathogen management and ultimately lead to higher crop production. Summary of the Invention [Problem to be solved by the invention]

[0005] As discussed above, there is a need to provide a reliable, efficient, computer-aided assessment of in-field soil-borne pathogens for target crops in an agricultural plot, and in particular nematode stress for soybean crops. [Means for solving the problem]

[0006] According to the present invention, this need is solved by a method for identifying soil-borne pathogens for a target crop in an agricultural land plot, as defined by the features of the independent claim 1. Preferred embodiments are the subject of the dependent claims.

[0007] In one aspect, the present invention relates to a method for identifying a soil-borne pathogen for a target crop on a farm plot, the method comprising: acquiring a first digital image of the farm plot during a first crop cycle, where a target crop is grown on the farm plot during the first crop cycle; acquiring a reference digital image of the farm plot during a reference crop cycle, where a reference crop is grown on the farm plot during the reference crop cycle, where the reference crop is different from the target crop; calculating a first vegetation index associated with a first pixel in the first digital image; determining a first signed distance between the first pixel and pixels surrounding the first pixel based on the first vegetation index; detecting a first anomaly for the first pixel if the first signed distance is less than a predefined threshold; defining a reference anomaly for the reference pixel of the reference digital image; and identifying a soil-borne pathogen for the first pixel if the first anomaly does not match the reference anomaly.

[0008] The method according to the invention may be realized as a computer program that processes digital images of the crop and detects soil-borne pathogens for the target crop with relatively little human intervention. In other words, the method relates to a computer-implemented invention, and the steps of the method described above and below may be implemented in a computer program. The computer program may be stored in a non-volatile memory. The computer may utilize its operating system and hardware components to execute the computer program for identifying soil-borne pathogens for the target crop in the agricultural land plot. The term "computer" may relate to a system including several computers or computing units in a distributed manner, as well as other electronic devices such as cameras and drones for acquiring digital images.

[0009] According to another aspect, the present invention relates to a system for identifying a soil-borne pathogen for a target crop on a farm plot, the system comprising an image capture device configured to acquire a first digital image of the farm plot at a first harvesting cycle and a reference digital image of the farm plot at a reference harvesting cycle. At the first harvesting cycle, the target crop grows on the farm plot. At the reference harvesting cycle, the target crop grows on the farm plot. The reference crop is different from the target crop. The system further comprises a calculation unit configured to calculate a first vegetation index associated with a first pixel in the first digital image, determine a first signed distance between the first pixel and pixels surrounding the first pixel based on the first vegetation index, and detect a first anomaly for the first pixel if the first signed distance is less than a defined threshold. The calculation unit is further configured to define a reference anomaly for the reference pixel of the reference digital image, and identify a soil-borne pathogen for the first pixel if the first anomaly does not match the reference anomaly.

[0010] The term "soil-borne pathogen" relates to a pest or disease that usually lives and reproduces in the soil and / or roots. It can cause damage to the root system and direct or indirect damage to the plant community. Soil-borne pathogens typically include fungi, bacteria, and nematodes. Crops in agricultural plots attacked or stressed by soil-borne pathogens, especially by nematodes, show symptoms such as stunting, yellowing and wilting of leaves. Such damage to the crop results in reduced yields. These symptoms to the community are recognizable by the human eye or by computer-based image processing means.

[0011] The term "digital image" in relation to a field plot relates to a digital image or photograph that is a visual photographic representation of the crops in the field plot. A digital image includes photographic elements, also called pixels. Each pixel is given a finite and discrete amount of a numerical representation of its intensity. If desired, the digital image in the representative invention may have a spatial resolution of 50 meters. In this case, one pixel of the digital image represents an area of ​​the crops that is 50 by 50 meters. For more advanced analysis, the spatial resolution may be finer, i.e., 20 meters or less. The digital image may be acquired, for example, by an airborne delivery platform such as a drone, a balloon, an airplane, or a spaceborne delivery platform such as a satellite. The digital image may be cropped using the boundaries of the field plot of interest, thereby selecting usable ones, e.g., potential cloud-free images.

[0012] The term "target crop" refers to the crop growing in the agricultural land plot that needs to be assessed. In other words, the target crop is a candidate crop on which the presence of a soil-borne pathogen is identified using the method of the present invention. In an exemplary embodiment, the target crop is soybean.

[0013] The term "reference crop" refers to a crop grown in a reference crop cycle. In order to compare a digital image of a target crop with a digital image of a reference crop, the reference crop is a crop different from the target crop. The reference crop cycle may be more recent than the target crop cycle. For example, to assess nematode pressure on a target crop in a field 12 months ago, a reference crop cycle from 6 months ago may be used, as long as it exists. Alternatively, the reference crop cycle may also be grown before the target crop cycle. In an exemplary embodiment, the reference crop is a non-host crop different from the target crop.

[0014] The term "vegetation index" is one of the indices for the observation and analysis of the earth's surface, more specifically the vegetation of the Earth. In relation to a pixel in a digital image, the vegetation index indicates the vigour of the agricultural land parcel represented by the pixel. A vegetation index of vegetation may be derived from a digital image of the agricultural land parcel. It relates to the vegetation characteristics of the agricultural land parcel. The vegetation index may be used to indicate the health of the vegetation of the agricultural land parcel, including the soil and the crops growing thereon. Typically, the Normalised Difference Vegetation Index (NDVI) is used as a benchmark, but other indices such as the Green Normalised Difference Vegetation Index (GNDVI) or single spectral bands may also be used.

[0015] The term "signed distance" quantifies how far the vegetation index of a given pixel deviates from the average vegetation index of the surrounding pixels, which is preferably NDVI. The signed distance may be calculated as defined by formula (1), where x is the vegetation index value of the central pixel being assessed, and μ and σ are the mean and standard deviation of the surrounding pixels.

number

[0016] "Surrounding pixels" refers to neighboring pixels that are not directly adjacent or next to the given pixel, but have a defined distance to the given pixel. For example, the defined distance may be at least 70 meters and at most 310 meters in terms of spatial distance within the agricultural land parcel, i.e., the neighboring pixels have a distance of 70-310 meters to the given pixel. From a geometrical point of view, the surrounding pixels form a donut-shaped window. Depending on the type of target crop, the defined distance may also be 80-310 meters, or 100-250 meters.

[0017] The term "abnormal" in relation to a pixel of a digital image refers to the state of the pixel that may have a soil-borne pathogen. For example, a pixel of a digital image under nematode stress is a pixel that is detected as abnormal during a first harvest cycle and is not detected as abnormal at all anywhere in the reference harvest cycle. A pixel under stress not related to nematode is a pixel that is detected as abnormal during the first harvest cycle of the subject and during the reference harvest cycle.

[0018] The present invention utilizes distinguishable image patterns to provide a reliable assessment of soil-borne pathogen stress in fields, such as nematode stress on target crops, such as soybean crops. The method according to the invention is based on remote sensing aerial or satellite imagery, making it scalable to large geographical areas. In particular, it relies on a time series of images capturing the rotation of a target crop, such as soybean, with a reference crop, such as a non-host plant, to separate above-ground soil-borne pathogen stress, such as nematode stress, from other non-nematode stresses. The underlying separation strategy is that nematode stress is only visible during the soybean cycle and not during the non-host plant cycle at the same specific location in the field, while other stresses are visible during both the soybean cycle and the non-host crop cycle. The same separation strategy applies to other crop-specific pathogens, such as Fusarium virguliforme (sudden death syndrome) on soybean or Plasmodiophora brassicae (clubroot) on oilseed rape.

[0019] Preferably, the reference anomaly may be defined by the steps of: calculating a reference vegetation index indicating the vegetation habitability of the agricultural land parcel for a reference pixel in the reference digital image; determining a reference signed distance between the reference pixel and the pixels surrounding the reference pixel; and detecting a reference anomaly for the reference pixel if the reference signed distance is less than a predefined threshold. Alternatively, the reference anomaly may be determined using conventional techniques, for example relying on human intervention. Furthermore, the reference anomaly may already be classified and stored in a historical database. In comparison, the present invention utilizes a computational technique for detecting the reference anomaly, whereby the reliability of the detection may be further improved. The detection of the reference anomaly using the above steps does not rely on the human eye, so that the reliability and effectiveness may be further improved. However, if the reference anomalies have already been classified using conventional techniques, such as the human eye, they may be used without reclassifying according to the above steps.

[0020] Preferably, the method includes a further step of using digital images of an additional crop cycle to refine the soil-borne pathogen identification result. The use of the additional crop includes the steps of acquiring a second digital image of the agricultural land plot in a second crop cycle, in which the target crop grows on the agricultural land plot in the second crop cycle, calculating a second vegetation index indicating the vegetation habitability of the agricultural land plot for a second pixel in the second digital image, determining a second signed distance between the second pixel and the pixels surrounding the second pixel, and detecting a second anomaly of the second pixel if the second signed distance is less than a defined threshold. In this way, the soil-borne pathogen identification result can be further improved in terms of reliability, for example increasing the accuracy of soil-borne pathogen detection of the target crop. By using images from an additional crop cycle, i.e. the second crop cycle, of the target crop, it is possible to further classify the nematode stress into two categories: (i) non-recurrent nematode stress, when the nematode stress is detected only on the first pixel and not on the second pixel, and (ii) recurrent nematode stress, when the nematode stress is detected on both the first pixel and the second pixel. Pursuant to the fact that nematodes can generally survive in soil for several years, recurrent nematode stress may also be considered as a more reliable classification.

[0021] Preferably, the first signed distance is determined by comparing the first vegetation index to the mean and standard deviation of the vegetation index of pixels surrounding the first pixel. As mentioned above, the first signed distance may be calculated using equation (1) or a threshold may be applied directly to the vegetation index, for example to isolate pixels with anomalously low vegetation index values ​​may also be used.

[0022] Preferably, the reference signed distance and the second signed distance may be calculated in the same manner as above, i.e., the reference signed distance may be determined by comparing the reference vegetation index with the mean and standard deviation of the vegetation index of the pixels surrounding the reference pixel. The second signed distance may be determined by comparing the second vegetation index with the mean and standard deviation of the vegetation index of the pixels surrounding the second pixel. As with the first signed distance, the reference signed distance and the second signed distance may be calculated using formula (1) or a respective threshold value may be directly applied to the respective vegetation index, for example, to isolate pixels with abnormally low vegetation index values ​​may also be used.

[0023] Preferably, the second harvest cycle may be used at any time before or after the first harvest cycle to aid in the identification of soil-borne pathogens in the target crop. Digital images of a plurality of second harvest cycles may be stored in a database, and the database may be searched if necessary.

[0024] Preferably, the method further includes generating a first signed distance map, a reference signed distance map and a second signed distance map comprising signed distances of a plurality of first pixels, a reference pixel and a second pixel, respectively.

[0025] Preferably, the first signed distance map, the reference signed distance map, and the second signed distance map may include a plurality of reference harvest cycles and a plurality of second harvest cycles, respectively. For example, an average reference signed distance map is calculated for each reference harvest cycle using a mathematical operator. For example, the signed distances of these multiple cycles may be merged by a minimum operator to highlight the most severe past stress. The signed distance map is also an image that visually represents anomalies or abnormal pixels. In particular, the signed distance map is a digital image with pixels, where pixel values ​​close to 0 represent normal pixels, pixel values ​​above 0 correspond to abnormally high vegetation index, and pixel values ​​below 0 correspond to abnormally low vegetation index. A further advantage of using the average signed distance map is that it reduces the impact of low signed distance occurrences on a single image. Low signed distances on a single image are seen to be due to noise, such as images with small undetected clouds or cloud shadows. In comparison, it is expected that nematode damage will result in vegetation stress that persists across multiple images.

[0026] Instead of using a signed distance map, any other method of isolating pixels with anomalously low vegetation index values ​​may also be used, including for example directly isolating anomalous pixels with vegetation index values ​​below a fixed threshold.

[0027] Preferably, the method further comprises the steps of calculating a target stress state for each of the plurality of target crops respectively using an average operator, and updating the target signed distance map by applying a minimum operator to the target stress state for each of the plurality of target crops. The step using the minimum operator is not essential. The step using the average operator may be required as there may still be multiple images within a target cycle.

[0028] Preferably, the method further comprises the steps of calculating a reference stress state for each of the plurality of reference crops respectively using an average operator, and updating the reference signed distance map by applying a minimum operator to the reference stress state for each of the plurality of reference crops.

[0029] Preferably, the method further comprises the steps of calculating a second stress state for each of the plurality of second crops respectively using an average operator, and updating the second signed distance map by applying a minimum operator to the second stress state for each of the plurality of second crops.

[0030] As mentioned above, low values ​​in the signed distance map correspond to abnormally low vegetation indices, which are holes in the vegetation. Thus, by using the minimum operator over multiple cycles, the most severe stress occurrences may be preserved, and using the average operator for a given harvest cycle is to try to reduce false stress detections due to noise. "Noise" here corresponds to something that leads to small signed distances on a single image, e.g. only an image with small clouds that were not filtered out. By comparison, nematode stress can be expected to persist over multiple images.

[0031] Preferably, the method further comprises identifying a soil borne pathogen for the first pixel if the first anomaly is not present in the reference signed distance map.

[0032] Preferably, the method further comprises identifying the first anomaly as a recurrent soil-borne pathogen if the first anomaly is present in the second signed distance map, or identifying the first anomaly as a non-recurrent soil-borne pathogen if the first anomaly is not present in the second signed distance map, In particular, if the anomaly occurs only during the first harvest cycle and not during the second harvest cycle, the anomaly may be classified as a non-recurrent nematode.

[0033] In other words, a more specific identification of soil-borne pathogens can be explained by using the comparison of the first signed distance map with the reference signed distance map and the second signed distance map. In particular, if the first anomaly does not exist in the reference signed distance map, the target crop may have soil-borne pathogens. Furthermore, if the first anomaly exists in the second signed distance map, it means that the target crop may have recurrent soil-borne pathogens, or if the first anomaly does not exist in the second signed distance map, it means that the target crop may have non-recurrent soil-borne pathogens. If the first anomaly exists in the reference signed distance map, it means that the target crop may not have soil-borne pathogens.

[0034] Preferably, the method further comprises the step of modifying the defined threshold for adjusting the level of soil-borne pathogens identified. Different thresholds may be useful for defining increased levels of soil-borne pathogen stress.

[0035] In particular, it can be used to isolate pixels that are below a certain value. Thus, the method may further comprise a step of isolating the first pixel with pixels that have a signed distance below a predefined threshold, the predefined threshold corresponding to a probability indicating a false alarm.

[0036] Preferably, the digital images have a spatial resolution of 50 meters or less, preferably 20 meters or less. The image resolution required will depend on the crop type and soil borne pathogen being assessed.

[0037] Preferably, each of the first vegetation index, the reference vegetation index and the second vegetation index includes at least one of a Normalized Difference Vegetation Index, a Green Normalized Difference Vegetation Index, and a Near Infrared band.

[0038] The present invention provides a method for detecting areas with above-ground stress due to soil-borne pathogens, such as nematode stress in soybean crops for a given geographical area. For example, it allows mapping soybean fields and identifying areas infested with nematodes. The method relies on a time series of remote sensing aerial or satellite images capturing the rotation of soybean with non-host plants to separate above-ground nematode stress from other non-nematode stresses. Non-nematode stresses may include soil compaction, nutrient deficiency, drought stress, and other non-soybean specific stresses. The method provides a reliable map of soybean nematode stress over a large geographical area. Such information is useful for better nematode management and ultimately for improving soybean production.

[0039] The method according to the invention is explained in detail below by way of an exemplary embodiment of the invention with reference to the attached drawings, in which: FIG. [Brief description of the drawings]

[0040] [Figure 1] FIG. 1 shows a flow diagram of the steps of an exemplary embodiment for identification of soybean regions under nematode stress. [Diagram 2] 1 shows an exemplary embodiment that relies on a time series of multispectral remote sensing images encompassing past cycles of soybean and non-host plants. [Diagram 3] 1 shows an exemplary embodiment of the calculation of NDVI for each digital image and then the signed distance for a donut-shaped sliding window. [Figure 4] 1 shows an exemplary embodiment in which a donut-shaped window is used to compare the pixel value of a central pixel with the mean and standard deviation of neighboring pixels. [Diagram 5] 1 illustrates an example embodiment in which a first signed distance map, a second signed distance map, and a reference signed distance map are converted into a first anomaly map, a second anomaly map, and a reference anomaly map, respectively, using a predefined threshold. [Figure 6]1 shows an exemplary embodiment of a signed distance map being combined with a map for the current soybean cycle, a map for past soybean cycles, and a map for past non-host plant cycles, the combination being done through average and minimum operators. [Figure 7] 13 shows an example embodiment in which a decision tree is used to determine if a pixel's stress during the current soybean cycle is nematode related or has another cause. [Figure 8] FIG. 13 shows an exemplary embodiment in which the combination of anomaly maps to an output shows areas under nematode stress with three stress intensities: high, medium, and low. [Figure 9] FIG. 13 shows an example embodiment in which the output of the method shows areas under nematode stress with three stress intensities: high, medium, and low, and drone images taken during the current soybean cycle show nematode plots highlighted on the output map. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0041] In an exemplary embodiment, the soil-borne pathogen is a nematode and the target crop is soybean. This is merely for better understanding and should not limit the scope of the claims directed to the general application of the method of identifying soil-borne pathogens to target crops.

[0042] In the following description, certain terms are used for convenience and are not intended to limit the invention. "Right", "left", "up", "down", "under" and "above" refer to directions in the drawings. The terminology includes the terms explicitly mentioned as well as their derivatives and terms with similar meanings. In addition, spatial relative terms such as "beneath", "below", "lower", "above", "upper", "proximal", "distal" and the like may be used to describe the relationship of one element or feature to another element or feature shown in the drawings. These spatial relative terms are intended to encompass different positions and orientations of the device during use or operation in addition to the positions and orientations shown in the drawings. For example, if a device in the drawings is inverted, elements described as "below" or "beneath" other elements or features will now be "above" or "over" the other elements or features. Thus, the exemplary term "below" can encompass both above and below positions and orientations. The device may be otherwise oriented (rotated 90 degrees or otherwise) and the spatially relative descriptors used herein interpreted accordingly. Similarly, descriptions of movement along and about various axes include various spatial device positions and orientations.

[0043] In order to avoid repetition of drawings and descriptions of various aspects and exemplary embodiments, it should be understood that many features are common to many aspects and embodiments. Omission of an aspect from a description or drawing does not imply that the aspect is missing from an embodiment incorporating that aspect. Instead, the aspect may be omitted for clarity and to avoid redundant description. In this context, the following applies to the remainder of this description. For clarity of the drawings, if a drawing contains a reference sign that is not described in the directly relevant part of the description, it refers to the previous or following description section. Furthermore, for clarity, if not all features of a part are given a reference sign in a drawing, it refers to other drawings showing the same part. Similar numbers in two or more drawings represent the same or similar elements.

[0044] The following is a detailed description of an exemplary embodiment of the present invention, namely, a method for reliable and scalable, within-field assessment of nematode impacts on soybean crops using remote sensing aerial or satellite imagery.

[0045] FIG. 1 shows a flow chart of some main and optional steps of an exemplary embodiment of the present invention, where the steps of assembling a signed distance map and isolating nematode stress are preferred but completely optional.

[0046] In the first step, a time series of multispectral remote sensing images are acquired. These digital images are visual photographic representations of the crops in the agricultural plots. In the second step, global stress can be detected by calculating signed distances for vegetation indices. In the third step, signed distance maps for each of the crop cycles can be assembled. In the fourth step, nematode stress can be isolated based on a decision tree applied to the stress maps.

[0047] In particular, the first step may involve preparing a time series of multispectral remote sensing images of the field to be analyzed. The images include spectral bands that allow for the calculation of vegetation indices that correlate to vegetation health. The Normalized Difference Vegetation Index (NDVI) is used here as the preferred index, but other indices such as the Green Normalized Difference Vegetation Index (GNDVI) or single spectral bands may also be used, as long as they emphasize the degree of vegetation habitability. The NDVI is defined as the normalized difference between the near infrared band and the red band using the following equation (2):

number

[0048] FIG. 2 shows a time series of images encompassing multiple crop cycles. Here, to assess nematode stress for the most recent soybean cycle, the time series encompasses three past cycles of soybean and three past cycles of non-host plants used in rotation with soybean. Some past cycles may not have cloud-free images available. This is acceptable as long as there is at least one past cycle for soybean and non-host plants with cloud-free images that provide a sufficient visual representation of the crop. Finally, all images may be cropped to the boundaries of the field to be analyzed and resampled on a common pixel grid to allow pixel comparison across images. This may be considered an optional pre-processing step, for example, the digital image in FIG. 2 shows only pixels within the plot. Pixels outside the plot are masked or blackened, so-called cropped to the field boundaries. Then, in the case of some satellite sources, or if different satellites are used, the images may be sampled on different pixel grids. Since a given pixel across multiple images is to be analyzed, the images should all be sampled on the same pixel grid. Recently, satellite imagery has tended to be provided in formats that allow for multi-temporal comparisons.

[0049] In this embodiment, there are three soybean crop cycles as the second crop cycle in the past, and three non-host crop cycles as the reference crop cycle in the past. These crop cycles are used to determine the nematode potential of the target crop, which is the current soybean, as shown in the very right place on the time scale. In other embodiments, the second crop cycle is more recent than the target crop cycle, and the reference crop cycle can also be more recent than the target crop cycle.

[0050] The second step, shown in Figure 1, may involve identifying field pixels that are under stress compared to neighboring pixels. First, NDVI is calculated for each image as a proxy to assess general vegetation health.

[0051] FIG. 3 shows a sliding window being applied to NDVI images to calculate the signed distance, defined by equation (1). The signed distance may be quantified by how far the NDVI of a given pixel deviates from the average NDVI of the neighboring pixels. This amount may be normalized by the standard deviation of the neighboring pixels to handle images with crops at different maturity levels. The window used to calculate the signed distance may have a donut shape, allowing pixels close to a given central pixel that may also be under stress to be ignored. The size of the window should not be smaller than the typical size of expected nematode stress, nor larger than other large scale field heterogeneities.

[0052] In particular, an NDVI image of the target crop including the first vegetation index may be generated using a digital photograph (digital image) of the target crop taken. Based on the NDVI image, a signed distance map may be calculated.

[0053] Figure 4 shows an exemplary donut-shaped window with an outer radius of 310 meters and an inner radius of 70 meters. The output of the second step is a set of signed distance maps, one for each image in the time series. The pixel being assessed is in the middle of the donut. Neighboring pixels adjacent or immediately adjacent to this pixel will be ignored from the analysis. Neighboring pixels surrounding the pixel being assessed, having a distance of 70 to 310 meters to the pixel being assessed, are used to calculate the vegetation index of the pixel.

[0054] FIG. 5 shows the first signed distance map (current soybean), the second signed distance map (past soybean), and the reference signed distance map (past non-host crop) in grayscale (top image) and their respective anomaly maps with abnormal pixels highlighted in gray, i.e., after transformation using a predefined threshold (bottom image). As mentioned above, low values ​​in the signed distance map correspond to abnormally low vegetation index, which is a hole in the vegetation. Accordingly, the three signed distance maps show abnormal pixels highlighted in black, which represent holes, and the three anomaly maps show abnormal pixels highlighted in (light) gray. An exemplary nematode section (np) on the first anomaly map is highlighted with a white arrow. On the reference anomaly map, the same area contains a non-nematode area (nn). In other words, the first anomaly map does not match the reference anomaly map in this exemplary area.

[0055] The third step shown in FIG. 1 may involve combining the multiple signed distance maps for each crop cycle into three maps: a map for the current soybean cycle, a map for past soybean cycles, and a map for past non-host plant cycles.

[0056] As shown in Figure 6, first, the average signed distance map is calculated for each crop cycle, and then the past cycles for a given crop are aggregated together using a pixel-wise minimum operator. The average operator captures the overall stress state of a given crop cycle and can handle a variable number of input images. The minimum operator separates the worst stress events of all past cycles. In this case, the non-host crop corresponds to the reference crop in the general definition of the method according to the present invention. Furthermore, in the general definition of the method according to the present invention, the past soybean corresponds to the second crop and the current soybean corresponds to the target crop.

[0057] The final step shown in FIG. 1 may involve combining the three signed distance maps through a decision tree to determine whether the stress of a given pixel during the current soybean cycle is nematode-related or has other causes, as shown in FIG. 6. A pixel is considered to be under stress if the signed distance value on each signed distance map is below a given threshold. In other words, a pixel is under stress if it corresponds to a negative NDVI anomaly. Different thresholds may be used to define an increased stress level. For example, from equation (1), a threshold of -2 separates pixels below -2σ. Assuming that non-stressed NDVI pixels follow a Gaussian statistical distribution, such a threshold corresponds to a probability of false alarm of 2.3%. Equation (3) gives the Probability of False Alarm (PFA) as a function of threshold t, mean μ, and standard deviation σ. The function erf is the error function.

number

[0058] FIG. 7 shows a decision tree of anomalies using signed distance pixels of the target crop cycle, which is the current soybean cycle. In particular, if an anomaly occurs during the first crop cycle and also occurs during the second crop cycle, the anomaly is classified as a recurrent nematode. Non-nematode-related stressed pixels are pixels that are detected as anomalies during the target first crop cycle and during the reference crop cycle. The first anomaly being a recurrent soil-borne pathogen means that the target crop has a recurrent soil-borne pathogen. The first anomaly being a non-recurrent soil-borne pathogen means that either (i) the target crop has a potential new soil-borne pathogen when the second target cycle is timely prior to the first crop cycle as shown in FIG. 7, or (ii) the target crop has a non-recurrent soil-borne pathogen when the second target cycle is more recent as the first crop cycle.

[0059] FIG. 8 shows an example of the output map of the method showing the combination of the first anomaly map, the second anomaly map, and the reference anomaly map using the decision tree according to FIG. 7, and the area under nematode stress. Three intensities of nematode stress are highlighted on the output map (low, medium, high). Areas under non-nematode stress are not shown, but are a beneficial by-product of the method of the present invention. Non-nematode stress may correspond to soil compaction, water stress, water logging, nutrient deficiency, and other non-soybean specific stresses. Here, the three stress severity levels "low", "medium", and "high" correspond to three thresholds t=-1.04 (15% PFA), t=-1.44 (7.5% PFA), and t=-2.33 (1% PFA), respectively. The white arrows point to the exemplary nematode plots (np) on the output map.

[0060] Overall, there are many advantages to using the method according to the invention, for example it can increase farm profits meaning higher yields and reduced inputs.

[0061] Furthermore, the method may identify locations in the field that are infested with nematodes. The nematode infestation map with GPS coordinates of the infested areas may be used for different purposes, such as (i) for direct targeted nematicide application, where treated seed or in-furrow nematicide applications are applied only to the infested areas, (ii) for targeted soil sampling to confirm the diagnosis and accurately assess nematode pressure in the infested areas, (iii) for crop rotation optimization, where a specific crop rotation may be selected for the whole field or parts thereof to reduce nematode pressure, especially in fields where high nematode pressure is expected, and (iv) to assess the evolution of nematode pressure in a field over time.

[0062] In addition, the method can be used to compare nematode pressure between different fields. For example, the method can identify the most infested fields in order to optimize and prioritize farm management. The evolution of nematode pressure over time in different fields can also be assessed to evaluate the effectiveness of applied countermeasures.

[0063] Finally, Figure 9 shows an aerial drone image on the right taken during the current soybean cycle showing the nematode plots (np) highlighted by arrows on the output map on the left, where areas under nematode stress are again shown at three stress intensities: high, medium, and low.

[0064] This description and the accompanying drawings illustrating aspects and embodiments of the present invention should not be taken as limiting the scope of the claims defining the protected invention. In other words, although the present invention has been illustrated and described in detail using examples such as nematodes and soybean crops, and standard signed distance maps in the drawings and the above description, such illustrations and descriptions should be considered as illustrative or exemplary rather than limiting. Various mechanical, compositional, structural, electrical, and operational changes may be made without departing from the spirit and scope of this description and claims. In some instances, well-known circuits, structures, and techniques have not been shown in detail so as not to obscure the present invention. It should therefore be understood that changes and modifications may be made by those skilled in the art within the scope and spirit of the following claims. In particular, the present invention encompasses further embodiments having any combination of features from different embodiments described above and below. For example, it is possible to operate the present invention in an embodiment in which a signed distance map is not used, as defined in claim 1 of the present invention.

[0065] The disclosure also includes all additional features individually shown in the drawings, which may not be described in the above or below description. Also, single alternatives of the embodiments described in the drawings and description, as well as single alternatives of those features, may be disclaimed from the subject matter of the present invention or from the subject matter disclosed. The disclosure includes subject matter consisting of the features defined in the claims or exemplary embodiments, and subject matter including the above features.

[0066] Moreover, in the claims, the word "comprising" does not exclude other elements or steps, and the indefinite articles "a" or "an" do not exclude a plurality. A single unit or step may fulfill the functions of several features recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Terms such as "essentially", "about", "approximately" and the like in connection with an attribute or value also specifically precisely define the attribute or precisely define the value, respectively. The term "about" in the context of a given calculable value or range refers to a value or range that is, for example, within 20%, within 10%, within 5%, or within 2% of the given value or range. Components that are described as being coupled or connected may be directly coupled, electrically or mechanically, or they may be indirectly coupled via one or more intermediate components. Any reference signs in the claims should not be interpreted as limiting the scope.

[0067] The computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, provided together with or as part of other hardware, but also distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. In particular, for example, the computer program may be a computer program product stored on a computer-readable medium, which computer program product may have computer-executable program code adapted to be executed to perform a particular method, such as the method according to the invention. Furthermore, the computer program may also be a data structure product or a signal for embodying a particular method, such as the method according to the invention.

Claims

1. A method for identifying soil-borne pathogens affecting target crops in agricultural plots, A step of acquiring a first digital image of the farm plot during a first harvest cycle, wherein the target crop grows in the farm plot during the first harvest cycle. A step of acquiring a reference digital image of the farm plot at a reference harvest cycle, wherein at the reference harvest cycle, a reference crop grows in the farm plot, and the reference crop is different from the target crop. A step of calculating a first vegetation index associated with a first pixel in the first digital image; a step of determining a first signed distance between the first pixel and surrounding pixels based on the first vegetation index; and a step of detecting a first anomaly in the first pixel if the first signed distance is less than a predefined threshold. The steps include defining a reference anomaly for a reference pixel of the reference digital image, If the first abnormality does not match the standard abnormality, the steps include identifying the soil-borne pathogen in the first pixel. Includes, A method wherein the target crop is a host crop, and the reference crop is a non-host crop.

2. The step of defining the aforementioned criterion abnormality is, The method according to claim 1, comprising the steps of: calculating a reference vegetation index related to a reference pixel in the reference digital image; determining a reference signed distance between the reference pixel and surrounding pixels; and detecting the reference anomaly of the reference pixel if the reference signed distance is less than the defined threshold.

3. A step of acquiring a second digital image of the farm plot during a second harvest cycle, wherein the target crop grows in the farm plot during the second harvest cycle. The steps include: calculating a second vegetation index indicating the vegetation viability of the farmland plot for a second pixel in the second digital image; determining a second signed distance between the second pixel and surrounding pixels; and detecting a second anomaly in the second pixel if the second signed distance is less than the defined threshold. The method according to claim 1, further comprising:

4. The method according to claim 1, further comprising the step of determining the first signed distance based on the first vegetation index and the mean and standard deviation of the vegetation indices of pixels surrounding the first pixel.

5. A step of determining a reference signed distance between a reference pixel and the surrounding pixels, based on a reference vegetation index indicating the degree of vegetation activity of a farm plot for a reference pixel in the reference digital image, and the mean and standard deviation of the vegetation indices of the surrounding pixels of the reference pixel, and / or The step of determining the second signed distance based on the second vegetation index and the mean and standard deviation of the vegetation indices of pixels surrounding the second pixel. The method according to claim 3, further comprising:

6. The method according to claim 3, further comprising the step of generating a first signed distance map, a reference signed distance map, and a second signed distance map, each including the signed distances of a plurality of first pixels, a reference pixel, and a second pixel, respectively.

7. The first signed distance map, the reference signed distance map, and the second signed distance map each include a plurality of first harvest cycles, reference harvest cycles, and second harvest cycles, The method according to claim 6, wherein the plurality of first harvest cycles, the standard harvest cycle, and the second harvest cycle alternate with each other.

8. A step of calculating the target stress state for each of the multiple target crops using an average operator, A step of calculating the baseline stress state for each of the multiple baseline crops using an average operator, The steps of updating the reference signed distance map by applying a minimum operator to each of the reference stress states of a plurality of reference crops, and / or The steps include: calculating the second stress state for each of several second crops using the average operator; Steps to update the second signed distance map by applying a minimum operator to the second stress state of each of the multiple second crops: The method according to claim 7, further comprising:

9. The method according to claim 6, further comprising the step of identifying the soil-borne pathogen of the first pixel of the target crop if the first anomaly is not present in the reference-signed distance map.

10. If the first anomaly is present on the second signed distance map, the step of identifying that the first anomaly is a recurrent soilborne pathogen, or Step 1: If the first anomaly is not present on the second signed distance map, identify that the first anomaly is a non-recurrent soil-borne pathogen. The method according to claim 9, further comprising:

11. The method according to claim 1, further comprising the step of modifying the predefined threshold for adjusting the level of the identified soil-borne pathogen.

12. The method according to claim 1, further comprising the step of separating the first pixels of the target crop by pixels having a signed distance less than the predefined threshold, wherein the predefined threshold corresponds to the probability of indicative a false alarm.

13. The method according to claim 1, wherein the soilborne pathogen is nematode stress, the target crop is soybean, and the reference crop is a non-host crop.

14. A non-volatile memory comprising a computer program for performing the steps described in any one of claims 1 to 13.

15. A system for identifying soil-borne pathogens affecting target crops in agricultural plots, An image capture device for acquiring a first digital image of the farm plot in a first harvest cycle and a reference digital image of the farm plot in a reference harvest cycle, wherein in the first harvest cycle, the target crop grows in the farm plot, and in the reference harvest cycle, the reference crop grows in the farm plot, and the reference crop is different from the target crop. The image capture device comprises the subject crop being the host crop and the reference crop being the non-host crop. A computing unit that calculates a first vegetation index associated with a first pixel in the first digital image, determines a first signed distance between the first pixel and surrounding pixels based on the first vegetation index, and detects a first anomaly in the first pixel if the first signed distance is less than a predefined threshold. Equipped with, A system in which the calculation unit further defines a reference anomaly for a reference pixel of the reference digital image, and identifies the soil-borne pathogen in the first pixel if the first anomaly does not match the reference anomaly.

Images