METHOD FOR DETERMINING THE DEVELOPMENT OF LEAF GREGATION

Description

technical field

[0001] The invention belongs to the field of plant cultivation, and more specifically relates to a method for determining the evolution of leaf wetness, that is to say, the appearance and disappearance of water on the surface of leaves of cultivated plants. Technological background

[0002] The presence or absence of water on the surface of cultivated plant leaves can lead to significant changes in a plant's health. In particular, prolonged water on the leaf surface can, when other conditions such as temperature are favorable, lead to the development of diseases, notably the growth of parasitic fungi. The presence of water on the leaves for a sufficient period is indeed the primary factor allowing the spores of fungal diseases (downy mildew, rust, powdery mildew, cercospora leaf spot, etc.) to germinate on the leaves of host plants (grapevines, roses, wheat, beets, etc.) and infect them.

[0003] The presence of water on leaves can be caused by rain, dew, or certain irrigation practices. When weather conditions are favorable to the development of fungal diseases, phytosanitary treatments are sometimes applied preventively, even though these treatments may not be necessary because the duration of water on the leaves was too short to allow for fungal infection. The inability to precisely measure the duration of plant wetness introduces uncertainty about the probability of infection, and to protect against fungal contamination, growers apply preventive treatments too frequently, whereas precise knowledge of the wetness duration would allow curative treatments to be triggered only when necessary.

[0004] The costs and environmental impact of plant protection treatments have led to a need to better define the conditions favorable to the development of parasitic fungi, foremost among them leaf wetness. Some existing methods for estimating the presence of water rely on imprecise overall data. For example, meteorological data measure relative humidity, temperature, atmospheric pressure, etc., and, based on physicochemical models, estimate the dew point, which is the temperature below which dew naturally condenses due to saturation. Leaf wetness is then determined solely in the case of dew. However, this is often an imprecise estimate, or one that requires numerous local measurements.

[0005] To improve accuracy, it has been proposed to use artificial leaves that emit a signal roughly proportional to the amount of water wetting them, for example, based on their surface impedance. However, this is an unreliable estimate, and above all, very localized, failing to take into account spatial variations such as terrain slope or the presence of a hedge.

[0006] Furthermore, leaf wetness can depend on factors other than dew, such as whether or not there has been rainfall, or even leaf drying, which is difficult to control because it depends on very localized characteristics like wind exposure and sunlight. Thus, simply knowing that it has rained does not tell us how long a leaf will remain dry. However, it is primarily the duration for which water remains on a leaf that determines the risk of fungal disease.

[0007] Finally, the hydrophobicity of the leaf surface (or of a fruit affected by a fungal disease, such as apple scab) impacts drying time and therefore wetting. An artificial leaf sensor, as described above, will only measure the presence of water on its surface, which may not have the same hydrophobicity as that of the plant, introducing a bias compared to the actual situation on the leaves.

[0008] The article PATEL ARTH ET AL: "Strawberry plant wetness detection using computer vision and deep learning", SMART AGRICULTURAL TECHNOLOGY, vol. 1, December 2021, page 100013, ISSN: 2772-3755, DOI: 10.1016 / j.atech.2021.100013 describes a technique for measuring leaf wetness by analyzing images in the visible and infrared spectrum, enabling the detection of water droplets on leaves and the deduction of the wetted surface area and the degree of leaf wetness. The article HEUSINKVELD ET AL: "A new remote optical wetness sensor and its applications", AGRICULTURAL AND FOREST METEOROLOGY, ELSEVIER, AMSTERDAM, NL, vol. 148, no.4, January 9, 2008, pages 580-591, ISSN: 0168-1923 describes a technique for determining leaf water content by measuring a ratio of light intensities at two wavelengths reflected by the leaves, the light beam having one wavelength selected to be absorbed by water and another wavelength selected not to be absorbed by water.

[0009] There is therefore a need for a means of determining the evolution of leaf wetness in an automated way, at positions spatially distributed in a cultivated area, with sufficient accuracy to best estimate the needs for intervention such as the application of a phytosanitary treatment. Presentation of the invention

[0010] The invention therefore aims to propose a method for determining the evolution of leaf wetness of a plurality of plants in a cultivated area which allows for a precise determination of leaf wetness at a distance from a detection system, and repeatedly over an observation period.

[0011] To this end, the invention proposes a method for determining the evolution of leaf wetness in a plurality of plants in a cultivated area, using a detection system comprising: a light source with an illumination field along an optical illumination axis, a light intensity sensor with an acquisition field along an optical capture axis, the light intensity sensor being configured to receive light radiation and to determine a light intensity signal, a data processing unit configured to receive the light intensity signal and determine a leaf wetness state, including the following steps: a) at each of a plurality of acquisition times, said acquisition times covering an observation period of at least 4 hours with intervals between acquisition times of less than 1 hour: a1) emission of a light beam by the light source towards a position of the cultivated area within its illumination field, a2) reception by the light intensity sensor of light radiation reflected from the position of the cultivated area, and determination of a light intensity signal from the reflected light radiation, the light intensity signal being a function of the quantity of light in the light beam reflected from the position of the cultivated area, b) for each light intensity signal of a plurality of light intensity signals from different acquisition times for the same position of the cultivated area, determination of a light intensity value for the position of the cultivated area from the light intensity signal,(c) determination of the light intensity value being representative of the light intensity of the light radiation reflected from the position of the cultivated area, based on light intensity values ​​at acquisition times for the position of the cultivated area, by comparing said light intensity values ​​to at least one reference value corresponding to the light intensity that would be reflected by dry leaves at the target position of the cultivated area when said leaves are illuminated by the light beam, resulting in an estimate of the leaf wetness state for the target position at the acquisition time.

[0012] Being able to measure the presence of water and the duration for which this water is present on plant leaves makes it possible to estimate the risks of spore germination, and therefore of contamination, and thus to apply appropriate control measures against fungal diseases.

[0013] The invention is advantageously complemented by the following various features taken individually or in their various possible combinations: plants have hydrophobic leaves, and hydrophobic leaves are considered wet when the light intensity value is greater than a reference value otherwise the hydrophobic leaves are considered dry; plants have hydrophilic leaves, and their wetness rate is estimated by comparing the evolution of the light intensity with a reference trajectory or curve measured during a calibration phase; a spatial distribution of leaf wetness is determined, for a plurality of positions distributed in the cultivated space;the light source and the light intensity sensor are part of an optical system configured to move the illumination field and the acquisition field, and step a) is repeated for a plurality of different positions of the cultivated space in its illumination field, the illumination field and the acquisition field being moved for each different position of the cultivated space; the light source and the light intensity sensor are part of an optical system comprising at least one reflecting element on an optical path of the light beam emitted by the light source and on an optical path of the light radiation reflected from the position of the cultivated space and received by the light intensity sensor; the reflecting element is movable and configured to move the optical axis of illumination and the optical axis of capture during an acquisition time or between two acquisition times;the light intensity sensor is an image sensor, and the light intensity signal is a two-dimensional image of the acquisition field; the light source is configured to emit a polarized light beam, and the detection system includes a splitter configured to separate the returned light radiation according to the polarization, and in which a first light intensity signal is obtained for a first polarization, and a second light intensity signal is obtained for a second polarization;The light source is configured to emit a light beam comprising several distinct wavelengths, and the detection system includes a separator configured to separate the returned light radiation according to wavelengths, and in which a first light intensity signal is obtained for a first wavelength range, and a second light intensity signal is obtained for a second wavelength range; the detection system includes a distance sensor configured to determine a distance between said system and the position of the cultivated space.

[0014] The invention also relates to a leaf moisture detection system comprising: a light source with an illumination field along an optical illumination axis, a light intensity sensor with an acquisition field along an optical capture axis, the light intensity sensor being configured to receive light radiation and to determine a light intensity signal, a data processing unit configured to receive the light intensity signal and determine a leaf wetness state, the system being configured to implement the process according to the invention. Presentation of the figures

[0015] Other features, purposes and advantages of the invention will become apparent from the following description, which is purely illustrative and not limiting, and which should be read in conjunction with the accompanying drawings on which: there figure 1 is a diagram showing an example of the steps involved in implementing a process according to a possible embodiment of the invention; the figure 2shows an example of a detection system implementing the invention according to one possible embodiment, with a composite light source, the figure 3 shows another example of a detection system implementing the invention according to one possible embodiment, with a combination of optical emission and capture paths, the figure 4 shows another example of a detection system implementing the invention according to a possible embodiment, with a light intensity sensor separating the received light according to a characteristic thereof.

[0016] The illustrated implementation methods can be combined with each other. Detailed description

[0017] Leaf wetness refers to the indication and / or quantification of the presence of water on a leaf. For example, leaf wetness can correspond to the "dry" or "wet" states of a leaf, with the wet state being further refined by degrees that quantify the presence of water or its distribution (density or droplet size, for example).

[0018] With reference to the attached figures, the method for determining the evolution of leaf wetness in a plurality of plants in a cultivated area uses a detection system 1 comprising: a light source 2 with an illumination field 4 along an optical illumination axis 6, configured to emit a light beam 8 a light intensity sensor 10 with an acquisition field 12 along an optical capture axis 14, the light intensity sensor 10 being configured to receive light radiation 16 and to determine a light intensity signal, and a data processing unit 20 configured to receive the light intensity signal and determine a leaf wetness state.

[0019] The light source 2 and the light intensity sensor 10 are part of an optical system 30 and are placed as close as possible to each other, typically a few centimeters or less. The optical system 30 is positioned near the cultivated plants so that they are within the illumination field 4 of the light source 2 and within the acquisition field 12 of the light intensity sensor 10. For example, the optical system 30 can be placed in a cultivated field or at its edge.

[0020] Preferably, this optical system 30 is raised above the leaves 22 of the cultivated plants and is typically mounted on a support such as a mast, pole, or tripod. Raising the optical system 30 increases the range of the illumination field 4 and the acquisition field 12, thus encompassing a larger area of ​​cultivated plants. This elevation also presents more of the upper surface of the leaves 22 to the optical system 30, and therefore within the illumination field 4 and the acquisition field 12. The elevation is preferably at least 1 m above the ground surface below the optical system 30, and preferably at least 1.5 m, and preferably at least 2 m, at the level of the lens through which the light beam 8 is emitted and / or through which the light 16 reflected by the plants enters.

[0021] Light source 2 can be a lamp, for example similar to a photographic flash, configured to emit intense light for a short period of time. Light source 2 can also be a laser.

[0022] The light source 2 can be configured to emit polarized light, for example, with linear, circular, or elliptical polarization. The light source 2 can be configured to emit white light, but preferably it is configured to emit monochromatic light. A wavelength of the light beam 8 emitted by the light source 2 is chosen to interact differently with the leaves 22 depending on the presence of water on them. Preferably, in order not to disturb wildlife, the light beam 8 is emitted in the infrared, with wavelengths greater than 780 nm. Several light sources 2 can be used, forming a composite light source 2 in which each light source 2 emits light with its own characteristics, for example, in wavelength and / or polarization.When several light sources are used, they preferably emit at distinct and different wavelengths. In the example of the... figure 2 , a first light source 2a can emit in a first wavelength, for example in the green, while the second light source 2b can emit in a second wavelength, for example in the red or infrared.

[0023] The light intensity sensor 10 can be a point photodetector, for example a photodiode. The light intensity signal is then directly related to the light intensity of the light radiation 16 received by the light intensity sensor 10, the origin of which is the light beam 8 emitted by the light source 2. The light intensity sensor 10 can also be an image sensor, and the light intensity signal is a two-dimensional image of the acquisition field 12.

[0024] The light source 2 emits light into an illumination field 4 along an optical illumination axis 6, while the light intensity sensor 10 captures the light radiation 16 into an acquisition field 12 along an optical capture axis 14. The optical illumination axis 6 and the optical capture axis 14 are close, forming an angle of less than 20°, and preferably less than 10°, and even more preferably less than 5°. To achieve this, the light source 2 and the light intensity sensor 10 are close to each other, and preferably housed in the same unit corresponding to the optical system 30.

[0025] Ideally, the optical illumination axis 6 and the optical capture axis 14 are coaxial and coincide at the output of the optical system 30. In particular, the light beam 8 and the light radiation 16 returned and captured by the light intensity sensor 10 can share the same portion of the optical path, and in particular pass through the same optical component of the optical system 30, such as the same optical lens.

[0026] To achieve this, the optical system 30 may include at least one reflective element 32 such as a mirror, a semi-reflective mirror, or any other optical component allowing the combination and / or separation of optical paths, onto which the light beam 8 and the light radiation 16 are reflected back to the light intensity sensor 10, as in the example of the figure 3. Such a reflective organ 32 is disposed on an optical path of the light beam 8 emitted by the light source 2 and on an optical path of the light radiation 16 returned from the position of the cultivated space and received by the light intensity sensor 10.

[0027] The optical system 30 is configured to move the illumination optical axis 6 and the capture optical axis 14, and therefore the illumination field 4 and the capture field. The movement can be achieved by moving the optical system 30, for example, by rotating the optical system 30. For example, as illustrated in the Figure 3, a reflective organ 34 such as a mirror disposed on an optical path of the light beam 8 emitted by the light source 2 and / or on an optical path of the reflected light radiation 16 is mobile, and by a change in its inclination moves the optical axis of illumination 6 and the optical axis of capture 14. It is then possible to carry out a scan of the positions of the cultivated space.

[0028] A measurement position is not necessarily a point, and can have a certain spatial extent. A measurement position can be as restricted as the area covered by a pixel of an image sensor, or even by adjacent pixels, but can also correspond to a plant, or even to the entire acquisition field.

[0029] The method includes an acquisition step implemented at multiple acquisition times, preferably for several locations within the cultivated area. The observation period covers an observation duration compatible with the duration of the measured leaf wetness phenomenon, and with intervals between acquisition times sufficient to accurately capture the evolution of this phenomenon. The acquisition times thus cover an observation period of at least 4 hours, with intervals between acquisition times of less than 1 hour. Preferably, the acquisition times are separated by intervals of less than 40 minutes, and even more preferably by intervals of less than 20 minutes. Preferably, the intervals between acquisition times for the same target location within the cultivated area are greater than one minute.For example, if a two-dimensional image of the entire acquisition field 12 is acquired at each acquisition time, this acquisition is repeated at each interval mentioned. If the acquisition involves scanning the acquisition field position by position within the cultivated area, then all the target positions are traversed during this interval. The interval between acquisition times is a compromise between the advantage of detecting leaf wetness with good temporal resolution and the inefficiency of excessively multiplying acquisition times, which would result in increased hardware requirements and large amounts of data requiring excessive processing time.

[0030] The acquisition times can be evenly distributed over the observation period, or they can have a temporal density that varies with the probability of the leaves being wet. For example, there may be more acquisition times when conditions are favorable for dew formation, namely, shortly before and after sunrise and sunset and during the night for dew, and during and after rain to measure the time of wetting due to rain. The observation period is typically at least 4 hours, but preferably more than 6 hours, and preferably also more than 10 hours. There are at least 4 acquisition times during the observation period, preferably at least 10 acquisition times, and preferably at least 20 acquisition times.

[0031] This acquisition step comprises two successive parts, which may, however, overlap temporally. In the first part (step S01), the light source 2 emits a light beam 8 towards a position in the cultivated space within its illumination field 4. Several light sources 2 can be used, preferably with coaxial optical illumination axes 6, for example with semi-reflective blades 3. The figure 2shows an example of the use of two light sources 2a, 2b forming a composite light source 2: a first light source 2a emits a first light beam 8 with first characteristics (wavelength, polarization), a second light source 2b emits a second light beam 8 with second characteristics (wavelength, polarization), and the first light beam 8 and the second light beam 8 are combined by a semi-reflecting mirror 3, thus giving a composite light beam 8 towards the position of the cultivated space.

[0032] After possible reflection from a reflecting element 34 of the optical system 30, the light beam 8 exits the optical system 30 and illuminates a location in the cultivated area where plant leaves 22 are situated. The leaves 22 may or may not have water on their surface. When the plants have hydrophobic leaves 22, the water forms more or less spherical droplets 24 on them. The droplets 24 and their support 22 behave as a reflective device (reflector or retroreflector) that reflects the incident light back in the same direction of incidence, i.e., towards the light source 2, regardless of the angle of incidence of the light rays with these droplets 24. This is the effect known as "heiligenschein". When the plants have hydrophilic leaves 22, this heiligenschein effect generally does not occur.On the contrary, the water on the surface of the leaves 22 tends to form a film 26 which diffuses the light, and therefore causes a reduction in the light intensity reflected back to the source, as in the example of the . figure 3 However, it is possible that even at low leaf wetness levels, droplets may appear on the surface of hydrophilic leaves, creating a short-lived Heiligenschein effect if dew continues to form. Therefore, the behavior of light intensity reflected by hydrophilic leaves during dew formation consists first of an increase in light reflection (as on hydrophobic leaves), then a transition to a decrease in reflection, and finally an intensity lower than that reflected by a dry leaf. Monitoring reflected intensity over time is thus important for assessing the leaf's wetness status.

[0033] Thus, part of the light beam 8 is reflected back to the light source 2 by the optical path taken by said light beam 8, and is therefore reflected back to the light intensity sensor 10. The intensity of the light radiation 16 reflected back to the light intensity sensor 10 therefore depends on the leaf wetness.

[0034] The light intensity sensor 10 receives (step SO2) the light radiation 16 reflected from the position of the cultivated area, and determines a light intensity signal from it. The light intensity signal is a function of the amount of light in the light beam 8 reflected from the position of the cultivated area, and therefore depends on the presence or absence of water on the leaves 22.

[0035] Preferably, the light intensity recorded by the light intensity sensor reflects the light intensity of the reflected radiation 16 in at least one wavelength in a range between 300 nm and 1350 nm, preferably between 300 nm and 1300 nm, and preferably also between 800 nm and 1200 nm so as not to disturb wildlife, or between 380 nm and 780 nm in order to use the visible spectrum.

[0036] In the example of the figure 2 The light intensity sensor 10 directly receives the reflected light radiation 16. In the example of the figure 3 The reflected light 16 travels back along the same optical path as the illumination beam 8, and in particular is reflected by a reflective element 34, then is deflected from the optical path of the illumination beam 8 by a semi-reflective mirror-type separator 32 to reach the light intensity sensor 10. In the example of the figure 4A separator 36 separates the reflected light 16 according to the polarization or wavelength of the reflected light 16: a first polarization or color is directed to a first light intensity sensor 10a, and a second polarization or color is directed to a second light intensity sensor 10b. It is thus possible to acquire a distinct light intensity signal for each polarization considered and / or for each wavelength considered. For example, a dichroic filter or a dichroic mirror can be used to separate wavelengths of the reflected light 16. Since the depolarization angle depends on the wavelength, it is advantageous to separate according to polarization, then according to color, or vice versa.

[0037] Light polarization can also be used to distinguish between the light reflected by water illuminated by the light beam 8 and another source reflecting the light beam 8, for example, a dry area of ​​a leaf. When the light beam 8 emitted by the light source 2 is polarized (elliptical, circular, or linear polarization), as is notably the case for a laser, the emitted light will tend to be depolarized by its reflection from the water present on the surface of a plant. Thus, by identifying, among the reflected light, the intensity of the light whose plane of polarization has rotated relative to the plane of emission of a light source such as a laser, for example, by means of a polarizing filter upstream of the light intensity sensor 10, it is possible to better highlight the light reflected by the water illuminated by the light beam 8.For example, it is possible to use a polarizing filter having a polarization that is inverse or complementary (e.g., perpendicular) to that of the light emitted by the light source 2. Additional filtering allowing the wavelength band of the light source to pass through when it is restricted, for example less than 50 nm, and preferably less than 20 nm, makes it possible to identify even better the only light radiation reflected by water illuminated by the light beam 8.

[0038] It is possible to perform this acquisition step several times in quick succession, for example within an interval of less than 10 seconds, for the same target position, in order to obtain several intermediate light intensity signals which are then combined (for example by averaging them) to give a light intensity signal for the target position at a given acquisition time. This approach makes it possible to compensate for certain noises, particularly when the intensity of the returned light radiation is low.

[0039] To compensate for ambient noise, especially when acquisition is done during the day, some (preferably all) of the acquisition steps may include a preliminary step of acquiring an ambient light intensity signal in the absence of the emission of a light beam 8. This ambient light intensity signal may be subtracted from the light intensity signal subsequently acquired, in order to subtract radiation not originating from the emission of the light beam 8.

[0040] Once this acquisition step is completed, and thus once a light intensity signal has been obtained for the target position in the cultivated area, the optical system 30 can move (step S04) the illumination optical axis 6 and the capture optical axis 14 to another position in the cultivated area, and a new acquisition step is implemented for this other position. It is therefore possible to scan a whole set of spatially distributed positions within the cultivated area. The distance at which it is possible to effectively target a position in the cultivated area depends primarily on the attenuation of the illumination light beam 8 and the sensitivity of the light intensity sensor 10. Typically, the target positions extend to a distance of at least 5 m from the optical system 30, and preferably at least 10 m from the optical system 30, and even more preferably at least 20 m.The sensitivity of the light intensity sensor 10 will allow the measurement distance to be increased further. The angular extension of the target positions around the optical system 30 depends on the ability to move the illumination and acquisition fields, and can even reach 360° if the optical system 30 is rotated, although an angular range of the order of 40 to 180° is more common for scanning by moving mirror, for example at the edge of a field.

[0041] The acquisition step is repeated for each targeted position in the cultivated area for several acquisition times covering the observation period. This yields, for each targeted position in the cultivated area, a sequence of light intensity signals from different acquisition times. For each light intensity signal in the plurality of light intensity signals from different acquisition times, a light intensity value is determined (step S05) from the light intensity signal. This light intensity value is representative of the light intensity of the light radiation 16 reflected from the position in the cultivated area. If the light intensity sensor 10 is a point sensor, such as a photodiode, the light intensity signal is directly a light intensity value.

[0042] It is also possible to consider moving the optical axis of illumination and / or the optical axis of capture during an acquisition period, i.e. with a continuous movement that sweeps a portion of the cultivated space.

[0043] When the light intensity sensor 10 is two-dimensional, as in the case of an imager or a plurality of photodiodes, the light intensity signal can include several representative values ​​of light intensity, such as several grayscale values, for the same position in the cultivated area. In this case, a light intensity value is derived for each position in the cultivated area. This can involve statistical calculations, for example, of a mean or median, possibly local, or simply an association between a value of the light intensity signal and a position. Thus, in the case of an image, the grayscale level of one or more pixels can be associated with a corresponding position in the cultivated area.In particular, when the light intensity sensor 10 is two-dimensional and acquires a two-dimensional image as a light intensity signal, the image comprises a two-dimensional set of pixels, each associated with a gray level. A light intensity value is then preferably an average value of gray levels from all or some of the pixels.

[0044] In the case where the optical system 30 moves the lighting field 4 or the acquisition field 12, an indexing of this movement makes it possible to associate each value of light intensity with a position in the cultivated space.

[0045] It is possible to plan for the optical illumination axis 6 and the optical capture axis 14 to remain fixed, and to always target the same position in the cultivated area. This is particularly feasible if the light intensity sensor 10 is an image sensor, and the light intensity signal is a two-dimensional image of the acquisition field 12, which associates each of a plurality of positions in the cultivated area with a respective light intensity value, typically a gray level.

[0046] To improve the localization of the target position, it is possible to equip the detection system with a distance sensor 9, for example of the Lidar type, and to acquire, at each acquisition time, a distance measurement associated with the target position, as in the example of the figure 2 Or 4This distance measurement allows for a more precise determination of the target position relative to the detection system, and thus enables a more accurate assignment of a light intensity value to a given position. Target positions can then be more easily organized spatially, for example, on a map. Knowing this distance also allows for the correction of distortions due to perspective. The role of the distance sensor 9 can also be played by the light source 2 when it includes several lasers, by using the "time of flight" of the light radiation 16 reflected back after the emission of the light beam 8 by the light source 2.

[0047] Separating the reflected light 16 according to wavelength or polarization allows for the production of several signals, with varying light intensity and depolarization, for the same target location. This makes it possible to better visualize the interaction between the light beam 8 and any water present on the leaves 22.

[0048] It is known that the polarization of a polarized ray incident on a water droplet is modified according to the shape and size of the droplet. This information thus allows us to infer, from the size of water droplets 24 on the surface of the leaves 22, or from the thickness of the film, the shape (dew or rain), and the density of these droplets 24 on the surface.

[0049] On the other hand, we also know that leaves absorb wavelengths that depend on their structure and composition. For example, they absorb red and blue light for photosynthesis, but reflect green light, hence their color, and also reflect near-infrared. The presence of water on the surface of the leaf can modify the relative fraction of each wavelength reflected by the leaf, but can also provide information about whether it is a leaf or another surface, such as soil or a stem. This allows us to provide contextual information about the type of surface struck by the incident beam, and therefore conclude that a reduction in the intensity of the reflected signal is due to the fact that it is soil rather than a leaf, and thus eliminate the measurement so that it does not bias the measurements on leaves.The interpretation of these different factors (depolarization, variation in light intensity, nature of the support) can partly involve the use of machine / statistical learning techniques, based on measurements taken on different known situations and which serve for possible supervised learning.

[0050] Leaf wetness is then determined at the position of the cultivated area based on the light intensity value associated with that position. For this purpose, the detection system 1 includes a data processing unit 20 configured to receive the light intensity signal and determine a leaf wetness state at the target position based on the light intensity signal. The data processing unit 20 includes at least one processor and memory, and communication interfaces for receiving data and communicating or interacting with a user. The data processing unit 20 can also be used to drive the optical system 30, causing the emission of the light beam 8 and the acquisition of the light intensity signal, or controlling the movement of the illumination field 4 and the acquisition field 12.

[0051] Determining leaf wetness can involve using a reference value, which typically corresponds to the light intensity that would be reflected by dry leaves 22 at the target location in the cultivated area when illuminated by the light beam 8. This reference value can be provided beforehand, or preferably obtained or updated in situ, for example by performing a calibration measurement when it is known that the leaves 22 are dry. These calibration measurements can be repeated at regular intervals (one to three times per week when the leaves 22 are dry) to avoid drift and to account for plant growth, which consequently alters the analyzed leaf area and the bare soil area.The device with a plurality of wavelengths can indeed measure the evolution of the leaf surface over time thanks, for example, to the calculation of the normalized difference vegetation index (better known by the English acronym NDVI for "normalized difference vegetation index") according to the choice of wavelengths.

[0052] In particular, the light intensity value can vary depending on the nature of the surface, but also on the distance between the target position and the optical system 30, because the power of the light beam 8 reaching the target position, and of the reflected light radiation 16, decreases with distance due to divergence or dispersion. The use of a distance sensor 9, such as a LIDAR, makes it possible, through the precise measurement of the distance between the target position and the optical system 30, to compensate for these distance effects, for example by modulating the exposure time of the light intensity sensor 10 for a given position, or by applying a correction factor to the measured values ​​or to the reference values ​​to which they are compared.

[0053] When plants have hydrophobic leaves, these leaves are considered wet when the light intensity value exceeds the reference value, due to the Heiligenschein effect mentioned earlier. In fact, the reflected light intensity is a monotonically increasing function of the wetness level. Preferably, the light intensity value is considered high if it exceeds 115%, and preferably 130%, of the reference value. Otherwise, the hydrophobic leaves are considered dry. Indeed, in the absence of water droplets on the leaves, the light from the beam is only slightly reflected because it is absorbed by the leaves.It should be noted that the light intensity increases with the density of drops 24 on the leaves 22 up to a plateau value, and the light intensity value therefore allows not only the presence of water on the leaves 22 to be determined, but also the quantification of this presence of water.

[0054] Conversely, when plants have hydrophilic leaves (22), the light intensity they reflect is a non-monotonic function of the leaf wetness level. Therefore, their wetness level can be estimated by comparing the evolution of light intensity with a reference trajectory or curve, for example, one measured during a calibration phase. For instance, using the reference value, hydrophilic leaves can be considered wet at that position when the light intensity value is lower than the reference value. Preferably, the light intensity value is considered lower if it is less than 90%, and preferably 80%, of the reference value, and even more preferably less than 70% of the reference value. As mentioned above, the Heiligenschein effect can, however, occur even with low degrees of leaf wetness in hydrophilic leaves (22).Thus, hydrophilic leaves can be considered wet at this position when the light intensity value is greater than the reference value. Preferably, the light intensity value is considered greater if it exceeds 115%, and preferably 130%, of the reference value. If the light intensity value is close to the reference value, and cannot be considered sufficiently lower or higher than the reference value corresponding to a dry leaf, the hydrophilic leaves are considered dry. It is also possible to consider a change in the light intensity value. For example, for a hydrophilic leaf, the appearance of dew may initially result in an increase in the light intensity value (appearance of droplets), followed by a decrease (appearance of the water film).Thus, the use of a reference curve makes it possible to take into account this temporal evolution of the reference value used to estimate the wetness state of the leaves 22.

[0055] The hydrophilic or hydrophobic nature of a plant leaf can be determined by the sphericity of water droplets on its surface. This sphericity is defined by the contact angle between the droplet and the leaf. Theoretically, this contact angle ranges from 0° for hydrophilic surfaces to 180° for hydrophobic surfaces. A surface is considered hydrophilic or hydrophobic when its contact angle is less than or greater than 90°, respectively. This angle depends on the surface free energy and the surface tension of the liquid. In plants, hydrophobic surfaces often contain hairs or a waxy coating. For example, rapeseed (Brassica napus) leaves have a surface wax that makes them hydrophobic, with a contact angle of approximately 110°–116°. Conversely, grapevine (Vitis vinifera) leaves are hydrophilic, with a contact angle of approximately 62° ± 14°.Thus, the more hydrophobic a surface is (e.g., the larger the contact angle), the more concentrated the backscattered beam will be. Theoretically, for an infinitely flat surface and a droplet resting directly on the surface, the maximum backscattering is reached at an angle of incidence of approximately 60° relative to the leaf surface. Under these conditions, the surface behind the droplet will be at its focal point. However, this value can be lower when considering the thickness of the surface (e.g., the focal point is within the surface) or a droplet raised above the surface (presence of transparent wax or hairs), and can, for example, be as low as 54°.

[0056] The light intensity values ​​recorded when the 22 leaves are dry can be used to update the reference value. For example, an average value can be derived from the light intensity values ​​closest to the previous reference value to provide a new reference value. Updating the reference value allows for adjustments to the growing area, including plant growth.

[0057] Leaf wetness can be determined by one or more comparisons resulting in an estimate of the leaf wetness state for the target location at the time of acquisition. Preferably, the estimate is a classification. Several reference values ​​can be used, each corresponding to a degree of wetness; the classification for obtaining leaf wetness is then based on the closest reference value. An important reference value, however, remains the light intensity value corresponding to the absence of water on the leaves.22

[0058] It is also possible to perform a regression (linear or non-linear) to obtain a curve where a light intensity value on the x-axis indicates the humidity level on the y-axis. We can therefore observe a continuum of humidity levels corresponding to a continuum of light intensity.

[0059] It is possible to use a neural network, for example a multilayer perceptron, previously trained to classify the degree of leaf wetness according to the light intensity value, a reference value being a value used during the training of the neural network.

[0060] Determining leaf wetness for multiple target locations can also be done globally for a portion of the cultivated area where these target locations are distributed. By classifying each target location according to wetness levels, a frequency analysis of these wetness levels can be performed to determine an overall wetness level. This is achieved by comparing the frequency of occurrence of light intensity values ​​corresponding to the presence of droplets to a reference frequency. The frequency can be spatial, but it can also be temporal, particularly when the acquisition and scanning of the target locations are performed continuously. For example, a reflecting device onto which the illumination beam, such as a laser beam, is incident can shift the illumination field to scan a portion of the cultivated area, such as the acquisition field.In this case, the reflecting element moves the illumination optical axis 6 and the capture optical axis 14 during an acquisition time. However, this acquisition time can be considered as encompassing acquisition moments, each corresponding to the target position at that acquisition moment. The light intensity sensor 10, for example a photodiode, determines a light intensity signal that varies temporally according to the movement of the illumination field based on the presence or absence of water at the scanned positions. By analyzing the frequency of the light intensity values ​​of the signal corresponding to the presence of water, it is possible to deduce an estimate of the wetness.

[0061] Since light intensity values ​​are available for different acquisition times at the same target location, it is possible to deduce the evolution of leaf wetness (step S06) over the observation period. It is then possible to determine when the leaves become wet and when they become dry, and therefore how long the leaves remain wet, which is the main factor allowing fungal disease spores to germinate and infect the leaves. It is then possible to determine the fungal risk for each target location, and consequently, it is possible to plan and carry out any necessary intervention, such as the application of a plant protection product, if required.

[0062] By having several target positions distributed in the cultivated area, a spatial distribution of leaf wetness can be determined, thus allowing a map of this evolution of leaf wetness, and therefore of the fungal risk, which makes it possible to intervene only in the areas that require such intervention.

[0063] The invention is not limited to the embodiments described and shown in the accompanying figures. Modifications remain possible, particularly with regard to the constitution of the various technical features or by substitution of technical equivalents, without departing from the scope of protection of the invention.

Claims

1. A method for determining changes in leaf moisture levels of a plurality of plants in a cultivated area, using a detection system (1) comprising: - a light source (2) with an illumination field (4) along an optical illumination axis (6), - a light intensity sensor (10) with a detection field (12) along a detection optical axis (14), the light intensity sensor (10) being configured to receive light radiation (16) and to determine a light intensity signal therefrom, - a data processing unit (20) configured to receive the light intensity signal and determine a leaf wetness state, comprising the step of: a) at each of a plurality of acquisition times, said acquisition times covering an observation period of at least 4 hours with intervals between acquisition times of less than 1 hour : a1) emission (S01) of a light beam (8) by the light source (2) towards a position in the cultivated area within its illumination field (4), a2) the light intensity sensor (10) receiving (S02) light radiation (16) reflected from the position of the cultivated area, and determining a light intensity signal (S03) from the reflected light radiation (16), the light intensity signal being a function of the amount of light of the light beam (8) reflected from the position of the cultivated area, characterised by the following steps: b) for each light intensity signal from a plurality of light intensity signals at different acquisition times for the same position in the cultivated area, determining a light intensity value (S05) for the position in the cultivated area from the light intensity signal, the light intensity value being representative of the light intensity of the light radiation (16) reflected from the position of the cultivated area ; c) determining a trend in leaf moisture (S06) for the position of the cultivated area over the observation period based on light intensity values at acquisition times for the position of the cultivated area, by comparing said light intensity values with at least a reference value corresponding to the light intensity that would be reflected by dry leaves (22) at the target position of the cultivated area when said leaves (22) are illuminated by the light beam (8), resulting in an estimate of the leaf moisture status for the target position at the time of acquisition.

2. A method according to claim 1, wherein the plants have hydrophobic leaves (22), and the hydrophobic leaves (22) are considered wet when the light intensity value exceeds a reference value, otherwise, the hydrophobic leaves (22) are considered dry.

3. A method according to any of the preceding claims, wherein the plants have hydrophilic leaves, and the hydrophilic leaves and their wetting rate are estimated by comparing the change in light intensity with a reference trajectory or curve.

4. A method according to any of the preceding claims, wherein a spatial distribution of leaf wetting is determined for a plurality of positions distributed throughout the cultivated area.

5. A method according to any of the preceding claims, wherein the light source (2) and the light intensity sensor (10) form part of an optical system (30) configured to move the illumination field (4) and the acquisition field (12), and step a) is repeated for a plurality of different positions of the cultivated area within its illumination field (4), the illumination field (4) and the acquisition field (12) being moved for each different position of the cultivated area.

6. A method according to any of the preceding claims, wherein the light source (2) and the light intensity sensor (10) form part of an optical system (30) comprising at least one reflecting element in an optical path of the light beam (8) emitted by the light source (2) and in the optical path of the light radiation (16) reflected from the cultivated area and received by the light intensity sensor (10).

7. A method according to the preceding claim, wherein the reflecting member is movable and configured to shift the illumination optical axis (6) and the capture optical axis (14) during an acquisition period or between two acquisition instants.

8. A method according to any of the preceding claims, in which the light intensity sensor (10) is an image sensor, and the light intensity signal is a two-dimensional image of the acquisition field (12).

9. A method according to any of the preceding claims, wherein the light source (2) is configured to emit a polarised light beam (8) and the detection system comprises a splitter configured to separate the returned light radiation (16) according to its polarisation.

10. A method according to the preceding claim, wherein a first light intensity signal is obtained for a first polarisation, and a second light intensity signal is obtained for a second polarisation.

11. A method according to any of the preceding claims, wherein the light source (2) is configured to emit a light beam (8) comprising a plurality of distinct wavelengths, and the detection system comprises a splitter configured to split the returned light radiation (16) according to wavelength, and wherein a first light intensity signal is obtained for a first wavelength range, and a second light intensity signal is obtained for a second wavelength range.

12. A method according to any of the preceding claims, wherein the detection system comprises a distance sensor (9) configured to determine a distance between said system and the position of the cultivated area.

13. A leaf moisture detection system comprising: - a light source (2) with an illumination field (4) along an optical illumination axis (6), - a light intensity sensor (10) with a detection field (12) along a detection optical axis (14), the light intensity sensor (10) being configured to receive light radiation (16) and to determine a light intensity signal therefrom, - a data processing unit configured to receive the light intensity signal and determine a leaf wetness state, the system being configured to implement the method according to any one of the preceding claims.