EP3369037A1

EP3369037A1 - Method and information system for detecting at least one plant planted on a field

Info

Publication number: EP3369037A1
Application number: EP16785383.7A
Authority: EP
Inventors: Roland-Kosta Tschakarow; Dejan PANGERCIC; Tobias DIPPER; Gabriel GAESSLER; Markus Hoeferlin; Tobias Mugele; Steffen Fuchs; Achim Winkler
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2015-10-28
Filing date: 2016-10-07
Publication date: 2018-09-05
Also published as: DE102015221085A1; US20180308229A1; WO2017071928A1; CN108140118A

Abstract

The invention relates to a method (300) for detecting at least one plant (110a, 110b) planted on a field (105). The method (300) has a step (310) of detecting a plant (110a, 110b) using an optical and/or infrared detection unit (120, 125, 130) in order to obtain plant image information (140) and detecting position information (160) which represents a geographical position where the plant (110a, 110b) is growing in the field (105). The method (300) further has a step (320) of identifying the plant (110a, 110b) using the plant image information (140) in order to obtain plant information (145) which represents the presence of the plant (110a, 110b). The method (300) finally has a step (330) of storing (330) the plant information (145) and the geographical position in a plant data set (165) in order to detect the plant (110a, 110b) planted on the field (105).

Description

Description title

Method and information system for recognizing at least one plant planted in a field

State of the art

The invention is based on a method for detecting at least one plant planted on a field or an information system as generically defined by the independent claims. The subject of the present invention is also a computer program.

In agriculture, the breeding of new varieties and in the

Optimization of Seed Quality applied largely standardized procedures for assessing breeding or treatment progress. These

Procedures are used in extensive series of experiments involving multiple repetitions in different locations.

The highest significance for breeders have test series (so-called series), which take place in the field and largely the actual

Reflect cultivation situation. For this field trial, therefore, several large-sized beds are created, which are divided into plots.

In most of the field trials Boniturverfahren are used, in which the characteristics of a plant to be assessed (germination, sprouting, stature height, leaf size, flowering, fruit ripeness, u.v.a.) accordingly

Development stage of a credit score (numerical value from 1 to 9). The characteristics to be considered depend on the type of crop. The assignment of the credit note numbers is done by human observers. The results are usually noted in the field on paper and transferred in a further processing step in a PC. In rare cases, today handheld computers (tablets) are used with specific input masks for the valuation of certain cultures and characteristics.

The data collected by persons in the field trial according to the rating scheme will be evaluated afterwards. The evaluation is usually a manual procedure in which the partial results are summarized in tables and then analyzed by means of statistical methods.

Disclosure of the invention

Against this background, with the approach presented here, a method for recognizing at least one plant planted on a field, furthermore an information system which uses this method, and finally a corresponding computer program according to the main claims are presented. The measures listed in the dependent claims are advantageous developments and improvements in the independent

Claim specified device possible.

The approach presented here provides a method for detecting at least one plant planted in a field, the method comprising the following steps:

Detecting a plant by means of an optical and / or infrared detection unit to obtain a plant image information and

Detecting position information representing a geographical position at which the plant grows in the field;

Identifying the plant using the plant image information to obtain plant information that represents the presence of the plant; and

Storing the plant information and the position information in a plant dataset to recognize the plant planted on the field. In the present case, a field can be understood as an acreage for plants or even a plot of such a field. For example, a crop can be understood as a crop whose fruit

used agriculturally, for example as food, feed or energy crop. By an optical detection unit, for example, a camera can be understood. For example, under a geographical position where the plant grows on the field

Information representing the position of the plant as a world coordinate or reflects the position of the plant with respect to a corner coordinate of the field. Plant image information can be understood to be an image or a plant parameter which reproduces optically detectable features or features of the plant which can be detected by infrared radiation. Here, the plant image information may also contain information obtained by processing or processing the image of the plant acquired by the optical camera and / or the infrared sensor. By identifying the plant, for example, determining the

Presence of the plant at the geographical position and / or a determination of a shape, size, species, number of leaves, leaf structure,

Number of buds, bud structure or other biological characteristics that makes a plant distinguishable from other plants. In this case, for example, it is also possible to determine a boundary of a plant in relation to another plant. Under one

In the present case, plant information can be understood as a parameter or information, such as the aforementioned shape, size, species, number of leaves, leaf structure, number of buds, bud structure or the like, which makes it possible to distinguish the plant from another plant. However, the parameter may also only provide information about the presence of the plant itself at the geographical position. A plant dataset may be understood to mean a data record or an information unit stored or to be stored in a memory in which the plant information is or will be stored with an associated geographical position of the plant.

The approach proposed here is based on the knowledge that under

Using a detection unit to recognize the plant certain geographical position can be made very quickly and reproducibly by avoiding a human classification. In this case, it is further exploited that, by storing the plant information and the geographical position in the plant data record, at the same time a machine-readable plant data record is generated that is suitable for a subsequent plant

Evaluation can be easily recycled. By assigning a geographical position to the plant or the plant information is further ensured that in repeated cycles of detection of the plant always the same plant can be detected, so that the

Developmental status of this plant can be traced clearly in time.

The approach proposed here offers the advantage that the assessment of the plant's plant growth can be carried out very quickly automatically. This is particularly useful when monitoring the growth of a large number of plants, as is the case for example in the

Seed research is required. Here, for example, in

different parts of the field different

Growth conditions created, for example by the application of different type or amount of fertilizer, this knowledge of the growth or environmental conditions in conjunction with the

Plant information and the geographical position then draw a conclusion on the influence of these growth or environmental conditions on the

Development of the plant at the corresponding geographical position. The approach presented here thus allows a significant improvement in the assessment of the developmental stages of growing in the field

Plant through an automated detection of the plant and the

automatically storing the plant information together with the geographical information in the plant dataset. According to an embodiment of the approach presented here, in the step of detecting the plant can be detected from a plurality of viewing directions and / or with an exposure time of less than half a second, in particular wherein the plant is detected with an exposure time of less than a tenth of a second. Such an embodiment of the here

presented approach offers the advantage of a very precise and possible to generate a defect-free image of the plant in order to exclude as far as possible the effects of disturbing the picture, for example by wind or shading of individual plant parts. Also favorable is an embodiment of the approach presented here, in which in the step of acquiring the plant image information using a previously read in plant distance information is detected, which represents a distance of plants in the field. Such a pre-read

Plant distance information can be, for example, information in which distance grid individual plants to be examined in particular

Plant of the field were originally planted. Such

Embodiment of the approach proposed here has the advantage that already a pre-selection of actually closer to be examined or to be detected plants can be made, which can sometimes save considerable effort in the recording or identification of plants.

Technically very simple and reliable is an embodiment of the approach presented here, in which in the step of identifying the plant using a color component of a plant image from the

Plant image information, in particular using a green

Color portion of the plant image from the plant image information and / or an infrared portion of the plant image of the plant image information is identified. Such an embodiment of the approach proposed here offers the advantage of the use of already mature algorithms

Image processing in order to be able to identify individual parts or whole plants in the plant image information.

Also, according to another embodiment of the approach proposed here, in the step of identifying, determination of growth information as partial information of the plant information, which includes a

Development status and / or or a health status of the plant represents, wherein in the step of storing the growth information as

Partial information of the plant information is stored in the plant dataset. Such growth information can, for example, information about the species, height, shape or number of leaves, buds, branches or the like or information about a structure of the plant itself or of parts of the plant. Such an embodiment of the approach proposed here offers the additional advantage that not only the presence of the plant, but already one or more other parameters which are responsible for the

Assessment of the state of development and / or health of the plant are recorded or stored in the plant dataset.

Also particularly advantageous is an embodiment of the approach proposed here, in which, in the step of identifying, a recognition of a species of the plant takes place in order to obtain plant information, in particular wherein recognition results in a distinction of the species of the plant from a species of another plant. A species of a plant may in the present case be understood as a genus of the plant. Such an embodiment of the approach proposed here offers the advantage of being able to make a distinction between a useful or a cultivated plant from a weed, which need not be taken into account, for example, for the investigation of the state of development of the plant. In this way it can also be established, for example, that at the geographical position where, for example, at an earlier time a crop plant was planted, but this has been received by drought or animal feeding, now one (not further)

considering) weeds plant grows. Such information that a weed plant now grows at the current geographic position may also be stored as part of the plant information in the plant dataset.

An embodiment of the approach proposed here, in which at least one further plant is detected by means of an optical and / or infrared detection unit in order to obtain at least one further plant image information and at least one is particularly advantageous for assessing the development state of a plurality of plants another geographical position is detected at which the further plant grows in the field. In the step of identifying, the further plant is identified using the further plant image information to obtain another To obtain plant information that represents the presence of the further plant. In the step of storing, the further plant information and the further geographical position are stored in the plant data record. Such an embodiment of the approach proposed here offers the advantage of being able to carry out the detection of a plurality, in particular of a multiplicity of plants, in a highly automated manner and thus quickly and reliably. In this way it becomes possible to get very accurate information about the

To develop the state of development of plants in the field.

Particularly advantageous is an embodiment of the approach presented here, which has a step of counting plants grown in the field using the plant information and the further plant information from the plant dataset. Such an embodiment of the approach presented here offers the advantage of a fast, safe and unambiguous counting of plants grown on the field.

Another advantage is an embodiment of the proposed here

Approach in which the steps of detection, identification and memory are carried out at least once repeatedly, in particular cyclically repeatedly executed, in particular wherein in each step of storing additionally a time parameter is stored, the one

Determination time of the plant information contained in the stored plant dataset represents. Such an embodiment of the approach proposed herein offers the advantage of providing a very precise history of the state of development of the plant.

Another advantage is an embodiment of the proposed here

Approach in which, in the step of acquiring, plant image information is acquired using information of a previously acquired or expected plant image information and / or in the step of

Identify the plant is identified using a previously recorded identification information. Such

Information of a previously recorded plant image information or identification information can be, for example, a recognition algorithm for the plant, which in a previous training cycle specifically for Detection of the plant or characteristics of the plant has been optimized. In this way, the detection or identification can be simplified, since the

Plant image information needs to be compared only with a reference information or a reference image and thereby usually much less numerical or circuit complexity to determine the

Plant information is required.

In order to be able to detect damage or contamination of the detection unit by simple means, for example, a plausibility check of the plant image information with a

Plausibilisierungsinformation from a temporally pre-recorded plant image information carried out to a functionality of the

Detecting detection unit. Such a time preceded

For example, a captured plant image may be plant image information taken in a previous embodiment of an embodiment of the method of recognition proposed herein. Such an approach offers the advantage of being able to detect a technical disturbance of the detection unit with simple means, so that as far as possible no erroneous data are stored in the plant data record and an early elimination of the disturbance at the detection unit is possible.

Particularly advantageous is an embodiment of the approach proposed here, in which, in the step of detecting, a detection unit arranged on a vehicle and / or on an aircraft is used to detect the plant. Such an embodiment offers the advantage of fast, unambiguous and reproducible detection of a large number of plants in the field.

This method can be implemented, for example, in software or hardware or in a mixed form of software and hardware, for example in a control unit.

The approach presented here also creates an information system that is designed to implement, control or implement the steps of a variant of a method presented here in corresponding devices. Also by this embodiment of the invention in the form of a device, the object underlying the invention can be solved quickly and efficiently.

For this purpose, the information system at least one arithmetic unit for

Processing of signals or data, at least one memory unit for storing signals or data, at least one interface to a sensor or an actuator for reading sensor signals from the sensor or for outputting data or control signals to the actuator and / or at least one communication interface to Reading or outputting data embedded in a communication protocol. The arithmetic unit may be, for example, a signal processor, a microcontroller or the like, wherein the memory unit may be a flash memory, an EPROM or a magnetic memory unit. The communication interface can be designed to read in or output data wirelessly and / or by line, wherein a communication interface, the

can read or output line-bound data, this data

for example, electrically or optically from a corresponding

Read data transmission line or in a corresponding

Can output data transmission line.

In the present case, an information system can be understood to mean an electrical device which processes sensor signals and outputs control and / or data signals in dependence thereon. The device may have an interface, which may be formed in hardware and / or software. In the case of a hardware-based embodiment, the interfaces can be part of a so-called system ASIC, for example, which contains a wide variety of functions of the device. However, it is also possible that the interfaces are their own integrated circuits or at least partially consist of discrete components. In a software training, the interfaces may be software modules that are present, for example, on a microcontroller in addition to other software modules.

In an advantageous embodiment, the information system controls an electronic record of the plant data record. For this The information system may, for example, read sensor signals such as the image data of the detection unit and an interface (for example from a satellite-based location system)

Access location information. The control takes place via actuators, such as a write head for an electromagnetic recording of the plant information and the geographical position information in the

Plant data set or in a microelectronic memory.

Also of advantage is a computer program product or computer program with program code which can be stored on a machine-readable carrier or storage medium such as a semiconductor memory, a hard disk memory or an optical memory and for carrying out, implementing and / or controlling the steps of the method according to one of the above

described embodiments, in particular when the program product or program is executed on a computer or a device.

Embodiments of the invention are illustrated in the drawings and explained in more detail in the following description. It shows:

Fig. 1 is a representation of a use of an embodiment of an information system for detecting at least one on a field

plant planted;

Fig. 2 is a schematic representation of an embodiment of the

Method of operating automated high-throughput plant counting and field surveying; and

3 is a flowchart of a method according to a

Embodiment.

In the following description of favorable embodiments of the present invention are for the in the various figures

represented and similar elements acting the same or similar Reference numeral used, wherein a repeated description of these elements is omitted.

1 shows an illustration of a use of an exemplary embodiment of an information system 100 for recognizing at least one plant 110 cultivated on a field 105. The plants 110 can thereby be arranged in columns, as shown in the left-hand area in FIG

for example, each be spaced in a planting distance 115 from each other. In this case, for example, when creating the field 105, the geographical position of the individual plants 110 may already have been determined and stored in a corresponding file. Now on the field 105

grown plants individually, the information system 100 has a detection unit 120 for detecting a (single) plant 110a. The detection unit 120 may include, for example, an optical camera 125 and / or an infrared sensor 130 (also referred to as an infrared camera) that form an optical image and / or an infrared image of the plant 110a. It is also conceivable that the detection unit 120 comprises a filter unit 135 in order to extract a color component, such as the green color component of the image captured by the optical camera 125. The optical image of the

Camera 125, an infrared image of the infrared sensor 130 and / or one of the optical image of the camera 125 / or an infrared image of the infrared sensor 130 derived image can then as plant image information 140 from the

Detecting unit 120 are output.

The plant image information 140 is then sent from a unit 145 to

Identifying read and determined using appropriate algorithms, the plant information 150. The plant information 150 represents the presence of the plant 110a. For example, the unit 145 may use an image processing algorithm for identifying this by using color components such as, for example, a green component in certain image regions, to draw a conclusion that a leaf or other component of the plant 110a would have to be imaged in these areas concerned. Additionally or alternatively, a

Edge detection algorithm used to outline the plant 110a and thereby recognize a spatial delimitation of the plant 110a of one or more adjacent plants.

Finally, the plant information 150 is transferred to a storage unit 155 into which the plant information 150 together with a

Position information 160, which represents a geographical position, is stored in a plant dataset 165. The position information 160 is in this case provided by a position detection unit 170, which receives the position information 160 using, for example, a

satellite positioning system 175, such as the GPS system. This position information 160 refers to the current geographical position, of which the detection unit 120 or a subunit of the detection unit 120 has created an image. In a larger spatial extent of the information system 100 can in this case in the

Position detection unit 170, where appropriate, an offset correction take place, which takes into account a distance of an antenna 180 of the position detection unit 170 of the image pick-up area of the detection unit 120.

It is also conceivable that also in the position detection unit 170, a file is stored, at which distance or at which geographical positions the plants 110 were planted, so that, for example, from the

Position detection unit 170 a trigger signal 185 is sent to the detection unit 120 when a geographical position is reached at which a plant 110a is expected. In this way, a numerical effort for the determination or evaluation of the of the detection unit 120th

recorded image or the recorded images are reduced because only at certain times of capture 120 images are taken, which are subsequently evaluated in the unit 135 and the unit 145 to identify. It is also conceivable that the position detection unit 170 also detects only a relative position of the information system 100 with respect to a corner coordinate 187 of the field 105 and thereby determines the relative position of the plant 110a. This relative position of the plant with respect to the corner coordinate 187 as a geographic position or as

Position information 160 is also sufficient here for the

subsequent evaluation of the information from the plant dataset to be able to extract relevant information regarding the particular plants such as the plant 110a. Furthermore, it is also conceivable that the position information 160 is supplied as a signal to the unit 145 for identification. In this case, unit 145 may already use the knowledge of the current geographical position from position information 160 to identify, for example, from previous recognition cycles a particular species of plant, plant size or the like by comparing the current plant image information with those contained in plant data set 165, from previously identified plant information. In this way, an effort for identifying the plant can be significantly reduced, since already some information about the expected species of PflanzellOa, size of the plant or the like can be suspected.

It is also conceivable that, when identifying the plant 110a also from the plant image information 140 further growth information is determined, for example, the size of the plant 110a, a number of leaves of the plant 110a, a number of buds of the plant 110a, a structure of the leaves of Plant 110a or other botanical features of plant 110a

from which a conclusion can be drawn on the state of development and / or the state of health of the plant 110a.

This growth information can then also be transmitted as part of the plant information 150 to the storage unit 155 and in the

Plant dataset 165 are stored. In this way, not only can a count of the individual plants 110 be made on the field 105, but it is also technically very easy to create a machine data record 165 which is very easy to machine and which provides a precise evaluation of the growth of the field 105 growing Plant 110 allows.

For this purpose, it is advantageous if not only a plant 110a is recognized using the above-described embodiment of the information system 100, but also other plants such as the plants 110b and 110c, for example, also recognized analogously to the procedure described above and depending on a corresponding combination of Plant information 150 with position information 160 for each of the plants 110b and 110c are stored in the plant dataset 165. It is particularly advantageous if, for example, according to the procedure described above, the information system 100 is used to determine a history of the growth of the plants 110 grown on the field 105. For this purpose, the information system 100 can be used at intervals, in particular cyclically, at intervals of days, weeks and / or months in order to detect the plants 110a, 110b and / or 110c cultivated on the field 105. In this case, after each detection of a plant 110, a corresponding combination of the plant information 150 with the respective position information 160 can be stored in the plant data record 165, wherein additionally a time information for this above combination of plant information 150 and position information 160 is stored, which stores a time, for example indicates a date and an hour to be entered by the detection unit 120

Plant image information of the corresponding plant 110a, 110b and / or 110c has been detected.

The recognition of the plants 110 planted on the field 105 is particularly simple if the information system 100 is fastened, for example, on or to a vehicle or aircraft not shown in FIG. As a result, detection of the plants 110 planted on the field 105 can be carried out quickly and reliably without major personnel expenditure, and a large and machine-evaluable plant data set 165 can be set up very quickly by technical means.

Compared to the procedure described above, however, the described manual Boniturverfahren are strongly subjectively shaped, because the results of Bonituren arise by consideration. Different observers may differ in their assessment in part strongly. In addition, the division into only nine qualitatively graded levels one

Constitute a limitation in the objective assessment of approximately equivalent candidates in breeding selection.

The approach proposed here is based on the automation of parts of the field trials described above. Be automated through data collection solved the disadvantages mentioned and at the same time achieved a higher throughput (field trials per unit time and personnel deployment).

Because of the manual compared to the above

Recording method significantly higher data storage capacity of the

Measuring computer arise further improvement aspects. The

computerized recording and management of the measurement data is up on

Single plant level possible and takes place directly on the field. In this

In connection with this, the history of the physiological development of each individual plant measured can be taken into account in the assessment of breeding candidates, varieties, seed qualities or pesticide use.

The approach proposed here can be used in particular for the assessment of the field emergence stage as an important development phase of crop plants.

In the field emergence stage, the task is the numerical and quota recording of rising plants. The field crop rate is an essential one, especially with increasing seed quality (combined with high seed costs) and precise application with precision seed drills

Quality feature (according to species and seed). The field count in the manual rating procedure is today a quota count. Consequently, a field emergence rate per plot is recorded. An assignment to the concrete individual plant is not possible in this way, which is why evaluation errors are possible even with correct counting (for example, germ plants disappear even after the casserole due to feeding or other external circumstances). The approach presented here overcomes this disadvantage because it allows the multiple (re-) recognition, counting and assignment of each individual plant in the stock to a unique position. Over the field emergence phase a history of the development of each plant can be determined at its assigned position.

At this stage, conclusions can be drawn from the actual leaf surface on the development progress of the individual plant in relation to the sowing or Earliest collection date to be drawn. One aspect of the approach presented here can therefore be seen in a possibility for reproducibly designed automatic recording of the leaf parameters of each individual seedling.

With the presented high-throughput measurement in the field taking into account each individual plant, the disadvantages of the manual Boniturverfahrens be overcome.

The correct plant count and recording of a position-related history is of great importance for the following objectives:

- in the assessment of seed quality in field trials (seed is offered in different qualities and for different

Site conditions optimized). A quality feature is the starting quota guaranteed by the manufacturer. Another quality-forming feature is the field emergence time. The simultaneous emergence of sowing results in a homogeneous stock. This is for plant care, for certain

Yield parameters and significant for the harvest. Homogeneous fields can be managed better (machine).

- When assessing breeding progress in field trials.

Breeding goals are z. As stress tolerances, resistance and fruit-specific characteristics. Yield is an integral breeding goal achieved through in conjunction with several specific subgoals.

- At the variety examination by institutional institutions. The of

Varieties developed by seed producers are tested and monitored in terms of their characteristics and characteristics by these institutions. The improved measurement and analysis technology allows testing to be performed faster, with greater accuracy, and less labor.

- In the development and use of plant protection products (PPPs). The function of plant count and position-related history recording can be used in conjunction with the use of PSM to control and further develop the performance of PSM. - To continuously improve knowledge about the suitability of varieties and seed qualities for the different locations worldwide

(Applications). For this, the methods according to the invention can also be used in daily agricultural operations. The data on field emergence and plant development gained during cultivation are inferior

Linking with climatic data and weather information in seedbed databases important indicators for the periodic suitability testing and improvement of site recommendations. Due to the progressive climatic changes, this is an important advantage for the seed producers and farmers.

The presented approach solves the disadvantages of the known and extends the applicability by additional advantages: a. ) Elimination of subjective influences by counting or measuring plants by persons b. ) Non-destructive high-throughput leaf area and height measurement of plants with the following characteristics:

- The single leaf surfaces of young and seedlings on single plant level

- total leaf area detection of individual plants c. ) Detection of the field emergence of each individual plant and assignment of a unique position (eg based on RTK GPS position, triangulation, trilateration, odometry, etc.) d. ) Recognition of the individual plant at the specific position and documentation of its physiological development on the basis of the measuring cycles based on the sowing and raising date e. ) Recognition of the loss of individual plants through feeding, drought, etc. f. ) High degree of automation of the measurements by using the measurement technology on vehicles / aircraft with integrated autonomous localization and navigation g. ) High degree of automation of the measuring processes by computer-aided recording and evaluation of large amounts of data h. The measurement is made by photogrammetric acquisition for the short duration of a single exposure. This excludes the influence of external factors such as B. Wind out. i. If several image recordings are taken one after the other (for example, while the measuring device is being moved over the inventory), then images of the corresponding plants in the image detail are obtained from different starting positions. The evaluation of the obtained images enables an improvement of the measurement results. j.) With the help of classification methods, the distinction

Data management software between examined crop and

Weeds. This classification is additionally supported by a-priori knowledge (such as alination, row and plant spacing). The cyclical measurements and classifications result in a refinement of the measurement and classification results, for example by transformation (in the process of alination) of the positionally determined position grid of the plants. k.) Due to the history of each plant contained in the data management in its original position, a correction or improvement of the

Data also retrospectively possible (for example, can

Misallocations are subsequently corrected).

I.) The data processing can both online (during the current

Data collection on the field) as well as offline (following to

Data recording). The online analysis also allows Plausibility checks of the acquired measurement data in order to minimize errors and external influences or exclude them on site. m.) The use of suitable methods for data compression and reduction allows the (longer-term) storage of measurement and analysis data on the device or a (radio) transmission to a cloud-based storage system. n.) Data Management also allows the collection and analysis of data collected by third-party devices or processes. Corresponding interfaces for data exchange are required. o.) From the obtained data further insights (eg over

Pest infestation, diseases, influence of external factors such as weather, water, drought) can be obtained. The connection to the

Influencing factors are of crucial importance for the assessment of breeding, seed treatment or plant protection measures. p.) The (semi-) automatic functions according to the invention contribute to a holistic examination of the plants in the herd (that is, as a result, the user obtains a consistent overall picture of the plants in the herd during different stages of development). q.) The invention takes into account a plausibility check on the basis of a (partial) online evaluation of the measured data. This ensures that faults in the measuring device (such as a soiled camera lens) are already detected at the beginning of the measurement and recording of the data. This ensures that no unsuccessful attempts and resources are spared.

The approach presented here includes two important elements: a. A metrological part of the functions Plant Counting and

Position assignment realized. b. A data management part that collects, archives, evaluates and / or processes the comparatively large amounts of data, for example, as meaningful reports and also helps to plan the measurements. Depending on the computing power and data connection, the recorded measurement data can be processed directly on the measuring instrument or transmitted to a central server. Alternatively, the storage on suitable data carriers is possible, which can be read in the wake of the measurement.

A suitable mobile solution is used by the measuring device (ie here

Information system 100) as a carrier platform for the execution of

Field trials. This mobile solution can be designed as a (semi-) autonomous vehicle, hand or motor vehicle, as an attachment for tractors, as an aircraft (for example as a drone) or in the simplest form in a portable form (for humans or animals).

Optionally, a plurality of parallel measuring devices 120 may be integrated in a structure for optimum construction of the measuring device 100

be able to use existing lanes or other conditions of the field structure meaningful.

In order to further parallelize tasks, it is also possible to distribute the measurements over several mobile devices (eg according to the swarm or master-slave principles). The here presented measuring device as

Information system 100 has a position and orientation sensor 170 (eg rtkGPS), optionally supplemented by an orientation sensor (eg an inertial measurement unit IMU), optionally supplemented by odometrically obtained position information.

It also has one or more cameras 125 with RGB, I R information channels and a suitable illumination unit for the required wavelength ranges. An optional shading unit of the image surfaces ensures that the disturbing influence of extraneous light and

Shading is minimized. The shading is designed so that it does not touch the plants when crossing. In detail, the following exemplary processes and functions are integrated into the measuring and evaluation device 100:

1. Given are the boundaries and dimensions of the total experimental area, the location and orientation of the plots and paths, and the trajectories resulting from the seed rows. These data are geo-referenced and recorded in a preprocessing step (eg when planting the trial plots or sowing). This geo-data are the measuring device 100 (here the information system 100) announced.

2. The concrete experimental design (here, for example, division of the parcels with their assignment to test objects and series) is likewise read into the measuring device 100 before the measurements are started (experiment description).

3. The measuring device 100 is guided in rows (eg in the drilling direction) over the test surfaces (that is to say here the field 105). This process can take place several times.

4. During the screening process (eg crossing) those in the

Measuring device 100 built-in cameras 125 and / or 130 of the

Detection unit 120 driven according to the known boundaries of the parcels and take data as plant image information 140. It is particularly noteworthy that one or more plants 110a and 110b are captured in the image section during the crossing. Thus, multiple images of the same plant 110a and 110b are sequentially made by the one-shot imaging technique (Image Sequencing). This method allows the influences of changing wind and light conditions as well

Overlaps and noise in the picture to a minimum.

5. The given by the cameras 125 and 130 under

For example, image data recorded in lighting conditions includes a.) RGB images and b.) Infrared images. The cameras 125 and 130 are calibrated (for example, to calculate the height assignment from the pictures). 6. The captured by the cameras 125 and 130 images are registered with each other to z. For example, an NDVI index (Normalized Differenced Vegetation Index, which is formed from reflection values in the near infrared and visible red wavelength range of the light spectrum) can then be calculated.

7. On the basis of the predetermined NDVI value, the biomass and soil are separated in the image, which results in biomass masks (reduced image data with clusterings of green fractions that are very likely to represent individual plants).

8. The biomass masks are assigned positions in the global coordinate system (eg rtkGPS positions). Here, several sensor data come into play in fusion (eg orientation of the measuring device, transformation between camera position and position sensors).

9. Based on the now georeferenced biomass masks are the

Classification method applied to the detected green clusters. In the case of a follow-on crossing, the biomass masks are re-identified on the basis of their global position data. The classification can be done online (during the current measurements) or offline (in the aftermath). The used classifiers have been trained for this purpose before the measurements. For this purpose, data has been collected on trial plots and manually examined and "labeled" (annotated) .The trainer distinguishes between crop and weeds in this process

Parameters such. For example, geometric features of the biomass masks (such as shape or size) and statistical features (such as the number of pixels of particular colors and their distribution based on intensity values). 10. The classification is based on all the image data recorded during the measurements (several images during the crossing, over several crossings). All classification results are taken into account and contribute to an overall result based on a majority voting evaluation process. This result is updated at the current date and can be used to correct misclassifications in the past correct. Overall, the method used increases the likelihood of correct classification.

11. In addition to the classification as described above, a priori knowledge about the location of the plant row as well as the spacing (eg planting distance) will be included in the final decision as to whether it is a crop or a weed.

12. Biomass masks classed as crops are used to calculate appropriate characteristics that contribute to the results of the trial (eg leaf area estimation, field crop detection or crop loss).

13. The data is stored after recording and evaluation z. B. stored in a cloud-based environment and maintained for visualization and further analysis on terminals. Evaluations can have different tabular, graphical and written form and contain all measured and derived parameters (eg scaled representation of the plants in their plots as plan view, weed stock in the trial field, etc.).

14. The data collected in different test series at different locations and over several years form a very large amount, which is usefully automated by the z. B. coupled via cloud services decentralized measuring devices are stored in a suitable database structure. For this purpose, the data must be reduced to the necessary and actually relevant level or additionally compressed. Suitable algorithms for searching and filtering allow individual plant, plot, serial, experimental and site-related as well as other cross-species and cross-cultural species

Evaluations.

As an example, the traceability and history of a

Single plant from cultivation to the consumer called. This also includes data obtained by the database outside the metering facility, such as: Geo-related information on weather, precipitation, soil values (temperature, conductivity, fertilizer concentration, etc.). Fig. 2 shows a schematic representation of an embodiment of the method with an operation of an automated high-throughput Pflanzenzählung and measurement in the field. In this case, in a carrier platform, such as the information system 100, the position detection unit is first used

170 is read a position information 160 representing the geographical position. This position information 160 can be detected, for example, via an interface such as an antenna 180 to a satellite-based location system 175, so that, for example, GPS data can be detected and thereby the geographic position 160 of the information system 100 can be determined. It is also conceivable that the position information 160 a memory in conjunction with odometrically detected movement data of the

Information system 100 is detected. Furthermore, the carrier platform or the information system 100 comprises at least one detection unit 120 for detecting a plant 110a or 110b or 110c by means of the optical camera

125 and / or the infrared camera 130 to obtain a plant image information 140 or an image derived therefrom. The detection unit 120 may, for example, comprise one or more sensor systems, such as the cameras 125 and 130, which are embodied, for example, as low-cost optical cameras which capture / capture images of the plants 110a or 110b and derive these as plant image information 140 or derived therefrom Information of unit 145 transmitted for identification. In the unit 145 for identifying, for example, a data extraction and / or a data fusion of the plant image information 140 and / or the information derived therefrom with the position information 160 can take place.

In this case, for example, as shown in greater detail in the region 210 of FIG. 2, an assignment 215 of the geographical position (which can be taken from the position information 160) to the plant 110a can be made. For this purpose, for example, the geographical position (for example using GPS data) of the plant 110a known from the position detection unit 170 and / or from a memory 220 can be used and / or used for plausibility checking.

Furthermore, a specific capture or identification 220 of the plant 110a may be in the form of the plant image information 140 or derived therefrom Information is provided so that, for example, this information in the

Plant data set 165 can be stored. This can be done in the

Plant data set 165 may be stored as part of the plant image information 140 or the information derived therefrom information about extracted features of the plant 110a, such as a sheet number, a leaf structure or the like of the respective plant 110a. Finally, in the storage unit 155, the plant information 145 together with the respectively corresponding position information 160 is stored in the plant data record 165.

In addition, it is conceivable that in the memory unit 155 or a corresponding control unit, not shown in FIG. 2, it is conceivable that

For example, exposure of the area covered by the camera (s) 125 or 130 is too weak, so that a not shown

Exposure unit (for example, as a subunit of the unit 120 for detecting) is driven in order to change the brightness in the area detected by the camera (s) 125 and 130, in particular to increase. Such

Drive signal 230 may be output to unit 120 for detection, for example, by unit 145 for identifying and / or storage unit 155 or the control unit (not shown in FIG. 2).

The plant dataset 165 may then, for example, to a

Processing unit 235 are output, either part of the

Information system 100 or, for example, in a central computer 237 is arranged, as shown by way of example in Figure 2. In this

Processing unit 235 then carries out, for example, the processing in a 4D processing level, as illustrated by way of example in region 240 of FIG. In this 4D processing level, an evaluation of the plant parameter data set 105, especially of the one preferably takes place

three-dimensional detected features of the plant 110 with the respectively associated

Times as the fourth dimension, so that a very precise inference to the growth or development of the respective plant 115 can be drawn from this information. Here are the in the

Plant data set 165 stored geographic positions 245, for example, as the GPS data of the plants 110a and 110b assigned Position information 160 is present, with data 250 relating to a history of the development of the individual plant (s) 110a or 110b or the re-identification of the individual plant (s) 110a or 110b, for example using the

Plant information 140 and / or the growth information, and the identified plant features 255 of the respective plant 110a or 110b linked, in order to determine from this example, the temporal growth of the individual plant 110a and 110b at the respectively associated geographical position. This allows, for example, with knowledge of

Soil, water content of the soil, weather and / or

Climate data at the geographical position of the field, assessment of the field

Environmental impact on the growth of the plant in the geographical position and thus, for example, to optimize the breeding of seeds. It is also conceivable that due to

Processing results of the executed in the processing unit 235 algorithms an annotation 238 takes place in which a change in the data of the stored in the storage unit 155 plant data set 165 takes place. In this case, for example, a correction can take place such that upon detection of a dead plant 110a or 110b, for example due to animal feeding or due to drought, the plant information for the plant 110a or 110b receives additional information such that this plant 110a or 110a resp.

110b is then no longer to be considered for a further assessment of the crops 110 grown in the field.

Furthermore, the plant data record 165 can be stored in a database 260, for example directly in the central computer 237, a PC or a cloud, not shown here. Also conceivable is a modification of the plant data record 165, which from the database 260 by means of a

Change signal 265 is triggered. Plant dataset 165 or results of a statistical evaluation of

Information from the plant dataset 165 can also be found in a

Display unit 265 are displayed visually. From this you can leave

Results 270 determine or determine the quality of a seed, a particularly favorable method of growing seed, a development of varieties of the plant, a development of the use of plant protection products etc. improve. Also conceivable here is a change of the

Plant data set 165, which is triggered by the display unit 265 by means of a change signal 275.

3 shows a flow diagram of an embodiment of the approach presented here as a method 300 for recognizing at least one plant planted on a field. The method 300 comprises a step 310 of detecting a plant by means of an optical and / or infrared detection unit in order to obtain plant image information and detecting position information representing a geographical position at which the plant grows in the field. Further, the method 300 includes a step 320 of identifying the plant using the plant

Plant image information to obtain plant information that represents the presence of the plant. Finally, the method 300 includes a step 330 of plant information and position information in a plant dataset to identify the plant planted in the field.

If an exemplary embodiment comprises a "and / or" link between a first feature and a second feature, then this is to be read so that the embodiment according to one embodiment, both the first feature and the second feature and according to another embodiment either only first feature or only the second feature.

Claims

claims

A method (300) for detecting at least one plant (110a, 110b) planted on a field (105), the method (300) comprising the steps of:

Detecting (310) a plant (110a, 110b) by means of an optical and / or infrared detection unit (120, 125, 130) to obtain plant image information (140) and detecting position information (160) representing a geographical position, where the plant (110a, 110b) grows in the field (105);

Identifying (320) the plant (110a, 110b) using the plant image information (140) to obtain plant information (145) representing the presence of the plant (110a, 110b); and

Storing (330) the plant information (145) and the position information (160) in a plant dataset (165) to detect the plant (110a, 110b) planted on the field (105).

2. Method (300) according to claim 1, characterized in that in

Step of detecting (310) the plant (110a, 110b) is detected from a plurality of viewing directions and / or with an exposure time of less than half a second, in particular wherein the plant (110a, 110b) with an exposure time of less than one tenth Second is detected.

3. Method (300) according to one of the preceding claims,

characterized in that, in the step of detecting (310), the plant image information (140) using a preamble

read plant distance information that represents a distance (115) of plants in the field (105). Method (300) according to one of the preceding claims, characterized in that in the step of identifying (320) the plant (110a, 110b) using a color component of a plant image from the plant image information (140), in particular using a green color component of the plant image the plant image information (140) and / or an infrared portion of the plant image is identified from the plant image information (140).

Method (300) according to one of the preceding claims, characterized in that, in the step of identifying (320), determining growth information as partial information of

Plant information (145) takes place, wherein the growth information represents a state of development and / or health of the plant (110a, 110b), wherein in the step of storing (330) the growth information as partial information of the plant information (145) stored in the plant dataset (165) becomes.

Method (300) according to one of the preceding claims, characterized in that in the step of identifying (320) a recognition of a species of the plant (110a, 110b) takes place in order to obtain plant information (145), in particular wherein a distinction is made in the recognition the species of the plant (110a, 110b) is from one species of another plant.

Method (300) according to one of the preceding claims, characterized in that in the step of detecting (310) at least one further plant (110b) is detected by means of an optical and / or infrared detection unit (120, 125, 130) to at least one obtaining further plant image information (140) and detecting at least one further position information (160) representing a further geographical position at which the further plant (110b) grows in the field (105), wherein in the step of identifying (320) another plant (110b) using the other

Plant image information (140) is identified to another

Plant information to preserve the presence of others Plant (110b) represents and wherein in the step of storing (330) the further plant information and the further position information (160) are stored in the plant dataset (165).

The method (300) according to one of the preceding claims, characterized by a step of counting plants (110, 110a, 110b) grown on the field (105) using the plant information (145) and the further plant information (145) from the plant dataset (165).

Method (300) according to one of the preceding claims, characterized in that the steps of detecting (310), identifying (320) and memory are carried out at least once repeatedly, in particular cyclically repeatedly, in particular wherein in each step of storing ( 330) additionally a time parameter is stored, the one

Determination time of the plant information contained in the stored plant data set (165) (145) represents.

The method (300) according to any one of the preceding claims, characterized in that in the step of detecting (310) the plant image information (140) using information of a temporally pre-recorded or expected

Plant image information (140) is detected and / or identified in the step of identifying (320) the plant (110a, 110b) using a previously recorded in advance identification information.

Method (300) according to one of the preceding claims, characterized in that in the step of detecting (310) a plausibility check of the plant image information (140) with a

Plausibility information with a time advance

recorded plant image information (140) to a

Functionality of the detection unit (120, 125, 130) to determine.

12. The method (300) according to one of the preceding claims, characterized in that in the step of detecting (310) arranged on a vehicle and / or on an aircraft

Detection unit (120, 125, 130) for detecting (310) of the plant (110a, 110b) is used.

13. Information system (100) that is set up to take steps of

Method (300) according to one of the preceding claims in corresponding units.

A computer program adapted to execute the method (300) according to one of the preceding claims 1 to 12.

15. Machine-readable storage medium on which the computer program

is stored according to claim 14.